Disertación/Tesis

Clique aqui para acessar os arquivos diretamente da Biblioteca Digital de Teses e Dissertações da UnB

2024
Disertaciones
1
  • Alex Alves Bernardes
  • Univariate and multivariate approaches to cardiac autonomic indices in predictive models of type 2 diabetes: analyses with statistical and machine learning techniques

  • Líder : FLAVIA MARIA GUERRA DE SOUSA ARANHA OLIVEIRA
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • FLAVIA MARIA GUERRA DE SOUSA ARANHA OLIVEIRA
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • TALÍA SIMÕES DOS SANTOS XIMENES
  • Data: 16-ago-2024


  • Resumen Espectáculo
  • Type 2 Diabetes Mellitus is a metabolic disease that commonly manifests in adults, often linked to an unhealthy diet and lack of physical activity. According to the World Health Organization, its incidence is continuously increasing worldwide. Diabetes Mellitus is highly associated with Autonomic Neuropathy (AN), a condition initially asymptomatic in the early years of diabetes diagnosis but significantly impacting the quality of life and increasing patient mortality. Early indicators of AN, even if the subject is still asymptomatic, can be used as predictors for the development of diabetes. The Autonomic Nervous System (ANS), part of the peripheral nervous system, regulates involuntary body actions such as cardiovascular function. This control influences heart rate, causing fluctuations between heartbeats, known as Heart Rate Variability (HRV). HRV is also influenced by the respiratory cycle and variations in blood pressure. It is possible to quantitatively assess the action of the ANS on cardiac regulation through physiological indices, measuring the impact of ANS dysfunction in diabetic individuals, with or without the presence of AN symptoms. Using this approach, the objective of this study is to determine quantitative ANS indices using univariate and multivariate techniques that can differentiate individuals with type 2 diabetes (T2DM) from control subjects. Additionally, the study aims to verify combinations of such indices that can predict diabetic individuals. These indices of autonomic function are determined from the processing and analysis of physiological signals such as electrocardiogram (ECG), continuous blood pressure, and respiratory signals. Initially, the study investigates the use of predictive models using indices of autonomic function determined from a single physiological signal, such as ECG or continuous blood pressure. Then predictive models using multivariate indices, that require data obtained from pairs of physiological signals, using both spectral techniques and system identification in the time domain as parameters for prediction, are developed. Data for this study was obtained from the public database “Cerebromicrovascular Disease in Elderly with Diabetes” (CDED). It is comprised of ECG, continuous blood pressure, and respiratory flow data from individuals aged 55 to 75 years, both with and without type 2 diabetes. The data were filtered and preprocessed using the CRSIDLab toolbox in MATLAB to extract autonomic indices. These indices were selected from the literature and categorized into three approaches: 1) statistical indices in the time domain, 2) spectral indices in the frequency domain, and 3) system identification indices in the time and frequency domains. Univariate time domain analyses provides statistical characteristics about a signal, such as mean, standard deviation, or event counts of interest. In the spectral domain, the estimation of power spectral density (PSD) of R-to-R time intervals (obtained from the ECG) and the time series of beat-to-beat systolic and/or diastolic blood pressure values (respectively the maximum peak and nadir in each heart beat interval) provide useful quantitative indices of autonomic function, based on the power density in different frequency regions of the PSD, and has been used in many studies to evaluate cardiac autonomic function in health and disease. However, since it is based on a single signal, other variables that may be related to the oscillations measured by the PSD in different frequency regions cannot be accounted for. The more specifically understand how the influence of different physiological signals, such as blood pressure and respiration, can cause changes in another signal, such as HRV, implies the need for an alternative approach that explicitly describes this tempora cause-and-effect behavior. To overcome the inherent limitations of univariate analyses, we also use system identification models in the time-domain to mathematically characterize the dynamic coupling between pairs of cardiovascular variables, using a causal impulse response (IR) model (in which the present output is restricted to depend on present and past, but not future, values of the input), from which additional indices of autonomic function can be obtained. All extracted indices of autonomic function were then analyzed using Pearson and Spearman correlations, and the One-way Anova statistical method. Correlation analyses were applied to verify the similarity between indices and the degree of influence of one index on another, while the One-way Anova technique was used to identify which indices are able to measure a statistical significant difference between diabetic and control groups. Subsequently, the indices were used as parameters in classification algorithms. The indices were first used individually and then in groupings. Different statistical and machine learning prediction models were used to quantify their performance in predicting T2DM vs. control subjects. Four machine learning algorithms were used in this study: 1) Support Vector Machines (SVM), 2) K-Nearest Neighbor (KNN), 3) Decision Tree (DT), and 4) Logistic Regression (LR). The different prediction models used indices that were grouped based on their signal origin and the extraction methodology used, as presented in Appendix B.1. The index groupings resulted in eight sets: A) HRV - statistical indices in the time domain and spectral indices in the frequency domain related exclusively to HRV; B) BPV - spectral indices in the frequency domain of Blood Pressure Variability (BPV); C) FRF - combination of frequency response function indices obtained from spectral analysis of HRV, BPV, and system identification by impulse response (IR); D) IR - system identification indices by impulse response in the time and frequency domains; E) HRV + BPV - the combination of indices in sets A and B; F) HRV + IR - the combination of indices in sets A and D; G) Special 1 - indices whose null hypothesis indicated a possible distinction between diabetic and control individuals with statistical significance; H) Special 2 - indices that achieved high performance in the prediction of diabetic and control individuals, present in all machine learning algorithms used. The classification of the test set based on individual indices showed excellent results when using indices calculated form multivariable models, such as the dynamic gain indices associated with cardiorespiratory coupling (CRC) or respiratory input of the coupling system with mutual influence of CRC and baroreflex coupling (BRC) were used. Following the respiration indices, the best indices for group prediction were the statistical indices of HRV combined with BPV, thus also a multivariate prediction model. The results with index groupings also showed that the use of HRV-related indices as model parameters can improve the accuracy, sensitivity, and precision performance of other marker sets in the classification and prediction of diabetic and control individuals. However, the grouping of indices related exclusively to HRV did not rank among the best performances in group prediction. The special groups 1 and 2 stood out as the best index groupings for predicting the studied groups of individuals, both containing statistical indices in the time domain and dynamic gain indices associated with the HRV-respiration pair of variables. Performance variation among the special groups occurred according to the machine learning algorithms used, where the groups alternated between the best results. The exception occurred for group G, which performed below the overall average when trained with the Decision Tree algorithm, which was led by grouping E. In all other algorithms, groupings G and H led the prediction performance. Finally, the special index groupings showed a slight improvement in group prediction compared to the prediction with individual indices using dynamic gain indices associated with respiration. This indicates a robust influence of respiration on HRV from a frequency perspective. It also highlights the importance of exploring index characteristics to avoid potential information conflicts when used as attributes for machine learning, as seen in the special groups.

2
  • Ricardo Gomes dos Reis
  • Development and Evaluation of a Video-Based Respiratory Management System and Distance Sensor for Radiotherapy in Voluntary Deep Inspiration Breath Hold in Patients with Left Breast Cancer

  • Líder : GLECIA VIRGOLINO DA SILVA LUZ
  • MIEMBROS DE LA BANCA :
  • GLECIA VIRGOLINO DA SILVA LUZ
  • MARILIA MIRANDA FORTE GOMES
  • RONNI GERALDO GOMES DE AMORIM
  • LEONARDO PERES DA SILVA
  • Data: 26-ago-2024


  • Resumen Espectáculo
  • ABSTRACT: In Brazil, breast cancer, after non-melanoma skin cancer, is the most common neoplasia type among women in all regions, accounting for approximately 30% of cases. Globally, according to the World Health Organization data, over 19 million new cancer cases were recorded in 2020, with approximately 2.26 million cases of breast cancer. Among the therapeutic modalities that enhance the survival of patients with this neoplasia is radiotherapy, capable of combating tumor growth through the use of ionizing radiation. However, its use is associated with cardiac and pulmonary toxicities, as it affects not only the diseased tissue but also adjacent healthy tissues. In this regard, some approaches have been developed to reduce doses, especially to the heart, during radiotherapy. The technique of voluntary deep inspiration has been shown to significantly reduce the average dose to the heart in patients undergoing radiotherapy for left breast cancer. However, commercially available devices for respiratory motion management are costly and not widely accessible in developing countries. This study proposes the development and evaluation of a low-cost respiratory management system based on two components (distance sensor and video camera), which could be used in radiotherapy for left breast cancer with voluntary deep inspiration. The functionality of the distance sensor was developed in the Arduino IDE and then connected to an algorithm developed for the video camera-based component in Python, using the Pyserial library. A user-friendly graphical interface was also developed in Python using the Easygui and OpenCV libraries. To assess the system, tests of accuracy, linearity, reproducibility, constancy, and response time were conducted. The system components were evaluated separately and in combination. Test results indicated that the distance sensor-based system component had accuracy below 0.5 mm, good linearity (R2=0.99), reproducibility less than 1.0 mm, and a response time of approximately 1.16 seconds. Favorable results were also found for the video-based component, with accuracy close to 0.1 mm and a response time of 0.12 seconds. The system's constancy test showed no risk of system locking during use. When both components were evaluated together, the system response time was 0.76 seconds. A comparison of the dose calculated by the planning system and the dose measured with an ionization chamber in a dynamic phantom (end-to-end test) showed a percentage dose variation between the planned and measured of 1.75%. Based on the results obtained, it is inferred that the developed system is capable of accurately mapping the variation in the position of the thoracic wall in breast cancer radiotherapy patients and can now be evaluated in clinical settings.

3
  • BRUNA CARVALHO FERNANDES
  • Electrophysiological evaluation of the perceptual processing of the Müller-Lyer visual illusion in healthy participants.

  • Líder : FABIO VIEGAS CAIXETA
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • FABIO VIEGAS CAIXETA
  • ISABELA ALMEIDA RAMOS
  • JAIR TRAPE GOULART
  • Data: 06-sep-2024


  • Resumen Espectáculo
  • This study explores the dynamics of visual perception through the use of the Müller-Lyer illusion and the techniques of Electroencephalography (EEG) and Event-Related Potentials (ERPs) to understand how different configurations of this illusion affect participants' perception. The methodology involved presenting several versions of the illusion to healthy individuals, measuring electrophysiological responses to identify specific patterns of visual processing. The results indicated possible significant differences in brain responses associated with variations of the illusion, highlighting the influence of geometric aspects on perceptual processing and the possible prevalence of bottom-up processing in the processing pathway of this illusion.

4
  • JOSE HENRIQUE BEZERRA CANDIDO
  • Psychophysical characterization of the perception of visual illusions in people with schizophrenia

  • Líder : FABIO VIEGAS CAIXETA
  • MIEMBROS DE LA BANCA :
  • FABIO VIEGAS CAIXETA
  • JAIR TRAPE GOULART
  • RUI DE MORAES JUNIOR
  • VALDIR FILGUEIRAS PESSOA
  • Data: 16-sep-2024


  • Resumen Espectáculo
  • Visual illusions have been used as tools for psychophysical tests in the investigation of visual processing deficits in many mental disorders. In the present study, we characterized the perception of schizophrenic patients in comparison to neurotypical subjects through subjective interpretations of the Müller-Lyer and Hollow-Mask illusions. The objectives included evaluating susceptibility to different illusions in both groups, highlighting possible cognitive deficits associated with schizophrenia, correlating responses to the Hollow-Mask illusion with medications used to treat schizophrenia, and exploring the potential use of visual illusions in diagnosing the disorder. Statistical analyses have shown a significant difference in susceptibility to the Müller-Lyer illusion between the schizophrenic patient group and the control group, suggesting altered perception in visuospatial judgment among the patients. Susceptibility was also tested across different angles in the Müller-Lyer illusion, revealing an inverse relationship between susceptibility and the widest opening angles. These results suggest that narrower angles may better differentiate schizophrenic patients from healthy subjects. Findings for the Hollow-Mask illusion diverged from those in the literature, as no significant difference was found between the groups. However, it was possible to associate the use of Olanzapine with resistance to the Mask illusion. We conclude that the interpretation of both illusions is influenced by perceptual and cognitive processes, and that schizophrenia may interfere with the integration mechanisms of these high-level processes, thereby supporting the potential diagnostic utility of visual illusions in identifying cognitive dysfunctions associated with this disorder.

2023
Disertaciones
1
  • Ana Paula Sandes de Souza
  • DEVELOPMENT OF A MULTIMODAL PHYSIOLOGICAL COLLECTION TOOL TROUGH COMMERCIALLY AVAILABLE ELECTROENCEPHALOGRAM AND EYE TRACKING DEVICES. CASE OF STUDY: BLINK DETECTION.

  • Líder : GERARDO ANTONIO IDROBO PIZO
  • MIEMBROS DE LA BANCA :
  • GERARDO ANTONIO IDROBO PIZO
  • MARILIA MIRANDA FORTE GOMES
  • JONES YUDI MORI ALVES DA SILVA
  • JAIRO JOSE MUNOZ CHAVEZ
  • Data: 27-ene-2023


  • Resumen Espectáculo
  • Electroencephalogram (EEG) and Eye Tracking (ET) are non-invasive ways to observe the behavior of the nervous system. The combination of these data types tends to have a higher classification capability than unimodal datasets when used in machine learning algorithms. Besides a higher accuracy, these datasets allow a better understanding of the physiological functioning by having more than one type of information. This work presents an alternative to clinical and high-cost equipment to build multimodal datasets. By using commercial equipment, this tool can generate a multimodal dataset of EEG and ET collected over time. The collection is mediated by MATLAB code, used to coordinate a pause-and-wait system between terminals connected in parallel to both hardware used. Subsequent data combination was performed in Python through resampling and linear interpolation, allowing unimodal data to be merged despite having distinct collection frequencies.

    The blink signal is recognized by both devices and was used to calculate the synchronization capacity between the signals. During collection, voluntary blinks occurred every 4 seconds. A total of twelve collections of 120 second duration were analyzed. A true positive blink identification rate of 93% and 99% were found for EEG and ET, respectively. The cross-correlation showed an average delay of 1.54 seconds of the ET signal relative to the EEG signal. Advances to improve the synchronization capability between the data were suggested.

2
  • Renata Menezes Lopes
  • Excitability modulation of descending pathways of the soleus muscle from the fatigue of contralateral plantar flexors

  • Líder : RINALDO ANDRE MEZZARANE
  • MIEMBROS DE LA BANCA :
  • FERNANDO HENRIQUE MAGALHÃES
  • JAKE CARVALHO DO CARMO
  • RINALDO ANDRE MEZZARANE
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • Data: 06-abr-2023


  • Resumen Espectáculo
  • Muscle fatigue (MF) is defined as a reduction in the capacity of the neuromuscular system to generate force, and is a common phenomenon in any type of activity that requires resistance. Responsible for limitations in human performance and causing injuries, the fatigue process can involve the ipsilateral limb as well as the contralateral limb, and can even be observed in non homologous limbs. Although the mechanisms of FM are exhaustively researched, few works have investigated the effect of contralateral fatigue (CF) on the lower limbs. In this context, aiming to expand the knowledge of CF, we propose to evaluate the modulation of excitability of descending pathways on the right soleus (SO) muscle after fatigue induction in the contralateral plantar flexors. The sample of the present study consisted of 12 healthy and active volunteers, 27.58 ± 7.34 years old, with a height of 1.67 ±0.08. Electromyographic (EMG) signals were captured by means of surface electrodes positioned on the right OS. A protocol was performed in which data regarding maximum voluntary contraction (MVC) and V-wave (electromyographic signal representing a response of the descending pathways) were collected in four conditions (pre-fatigue, between fatigue, post-fatigue and recovery). V-wave, power spectrum parameters of the contralateral OS electromyogram and in superimposed twitch (ST) force generation were obtained, such as mean power frequency (FMD), median power frequency (FMN) and Root Mean Squared (RMS) between conditions. A one-way repeated measures ANOVA was used to detect differences between the amplitudes of evoked V-waves, in FMD and FMN, in RMS across conditions. The main results of this study were a) that fatigue induced in the left ankle plantar flexors was shown to have a significant effect of CF on V-wave amplitudes; b) a tendency to recover from the effects of fatigue over time; c) modulation of the V-wave by CF was progressive; d) the absence of ipsilateral fatigue; e) that voluntary muscle activation was constant across conditions; f) that V-wave fluctuations were not attributed to different levels of voluntary activity of the target muscle. The present work evidenced that the excitability of descending pathways to the OS muscle can be acutely modulated as a consequence of fatigue induced in the contralateral plantar flexor muscles.

3
  • Jonathan Martinichen
  • Clinical Engineering Teams in Brazil: A Profile Analysis from the Perspective of Engineers who Work in this Field

  • Líder : MARILIA MIRANDA FORTE GOMES
  • MIEMBROS DE LA BANCA :
  • GERARDO ANTONIO IDROBO PIZO
  • MARILIA MIRANDA FORTE GOMES
  • RAFAEL FONTES SOUTO
  • RONNI GERALDO GOMES DE AMORIM
  • Data: 27-abr-2023


  • Resumen Espectáculo
  • Healthcare Establishments (HE), in order to guarantee the quality and safety of Medical/Hospital Equipment (MHE), as well as reduce acquisition and maintenance costs, must have their own, outsourced or mixed Clinical Engineering (CE) team. This team is responsible for preparing and executing plans for the acquisition and execution of MHE and, as in other areas governed by quality in management, the results may vary according to each HE and its CE maintenance teams, enabling to monitor the efficiency and effectiveness of provided services. This study aims to present the profile of CE teams and analyze how performance indicators relate to the composition of CE sectors and the characteristics of HE. As far as methodological tools are concerned, questionnaires were sent to clinical engineers in all units of the Federation, belonging to a WhatsApp group called “Engenharia Clínica Brasil”. The questionnaire gathers information about the engineers' education, the characteristics of the EAS, the profile of the EC teams, as well as quality indicators related to maintenance activities. The findings suggest that the composition of CE teams and the characteristics of HE may have a significant impact on the effectiveness and efficiency of maintenance activities. Therefore, it is crucial for HE to monitor the performance of their CE teams and take necessary actions to ensure the quality and safety of medical equipment, while minimizing acquisition and maintenance costs.

4
  • SAMYLA DE SOUZA MELO
  • HEALTH TECHNOLOGY PROPOSAL FOR TREATMENT OF PRESSURE INJURIES: RAPHA® EQUIPMENT

  • Líder : MARIO FABRICIO FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • JOSÉ CARLOS TATMATSU ROCHA
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • MARIO FABRICIO FLEURY ROSA
  • SOLANGE BARALDI
  • Data: 28-abr-2023


  • Resumen Espectáculo
  • Introduction: Pressure injury is defined as a localized injury to the skin or underlying cutaneous tissue over a bony prominence, resulting from a combination of pressure and local shear. It is caused by the sum of clinical factors, such as nutrition, hydration, and cutaneous-mucous conditions, among others. It occurs in several regions of the body, however, the region that has a greater shear is the sacral region. According to (PATTON, 2018), the use of new technologies decreases the prevalence of pressure injuries associated with mortality rates in a study developed evaluating the technologies used in pressure injuries in hospitalized patients. Based on the results of the clinical trial on the Rapha Equipment research (CAAE 94910718.5.3001.5553) where it prioritized the application of the protocol linked to the LED light-emitting equipment associated with a latex dressing (Rapha Equipment) (FLEURY ROSA, S.S.R, et al, 2017) for the treatment of foot diseases in diabetes, the following proposal was thought: what clinical outcomes would Rapha equipment provide in patients with pressure ulcers? Objectives: The objective was to evaluate the action of the LED light-emitting device together with the latex sheet (Rapha® Equipment) applied to a patient with a pressure injury hospitalized in a hospital institution during a 45-day protocol. Observing at the end of the research if there was attenuation of the total area of the lesion after the protocol. Materials and methods: A multi-strategy typology approach supported by quantitative and qualitative methods for data collection and analysis with an interdisciplinary health perspective, using the approved clinical study research (CAAE 94910718.5.3001.5553). The application methodology was similar to applying a pressure sore dressing, following the predefined protocol for applying the Rapha® equipment daily on the wound for 45 days. In each patient, it was evaluated whether the use of the Rapha® equipment resulted in a reduction in the area of the lesion. Results: Five patients from the experimental group who received the use of the Rapha® equipment and 4 patients from the control group who received the sus protocol were evaluated. Patients in the experimental group had two completely healed lesions and two lesions with a 50% decrease in the total area of the lesion. Compared to the control group, there was no healing and a small percentage of reduction in the area of the lesion. Conclusion: The Rapha® equipment assists in the healing process of pressure injuries in less time than the Unified Health System protocols, however, it favors increased moisture in the injury, which in excess can harm the healing process.

5
  • Ana Luiza Moraes Fernandes da Costa
  • IDENTIFICATION OF LARYNGEAL LESIONS FROM NARROWBAND ENDOSCOPY IMAGING USING ARTIFICIAL NEURAL NETWORKS AND VISUAL PROGRAMMING 

  • Líder : GERARDO ANTONIO IDROBO PIZO
  • MIEMBROS DE LA BANCA :
  • GERARDO ANTONIO IDROBO PIZO
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • RENATO CORAL SAMPAIO
  • LUCIANA CORREIA ALVES
  • Data: 08-may-2023


  • Resumen Espectáculo
  • The presence of certain types of lesions in the laryngeal mucosa can signal the development of early-stage laryngeal squamous cell carcinoma, which arises from the abnormal development of the squamous cells that make up the inner layer of the laryngeal lining. This disease corresponds to 98% of the occurrences of malignant tumors that affect the larynx and 3% of all cases of cancer, often leading to death when diagnosed in advanced stages.  

    Early detection of lesions suggestive of the development of precancerous tissue or early stage laryngeal cancer can be a challenging task even for experienced physicians. The existing scientific literature points out that models based on machine learning have been used to aid the diagnosis of laryngeal cancer, achieving relevant results, considering the state-of-the-art architectures of deep convolutional artificial neural networks in the detection of laryngeal lesions, with reported accuracy of up to 98%. However, the implementation of machine learning models often requires advanced knowledge in programming language, making the task of training and developing these solutions complex. There are also more recent models of artificial neural networks, not yet explored for the classification of images containing laryngeal lesions.  

    The present study aimed to develop methods based on artificial intelligence to classify laryngeal lesions in an automated way, from digital images obtained by the narrowband imaging technique.  

    Regarding the adopted methodology, 1.320 digital images of laryngeal tissue, healthy and in an early stage of cancer, obtained through laryngoscopy, were used, subdivided into 4 classes: healthy tissue, tissue with leukoplakia, tissue with hypertrophic vessels and tissue with capillary loops intrapapillary. From this set, 132(10%) were segregated to compose the test set. Five different machine learning models, based on deep convolutional neural networks (CNN) and capsule networks (CapsNet), were implemented to classify the 132 images: VGG16, VGG19, Inception V3, CapsNet without data augmentation and CapsNet combined with a data augmentation technique using images generated synthetically by an adversarial generative network (GAN).  

    The neural network architectures were implemented using the free software ORANGE Data mining and the Google Colab computational platform, in Python language, using the Keras, OpenCV and Tensorflow libraries.  

    Using GAN to augment the training set with synthetically generated images improved the performance of the capsule classifier in classifying all types of injuries. The CapsNet classifier with data increase by GAN reached average recall,precision and F1-Score of 94.7%, reaching the second best performance among the studied models. The best performance was obtained with the CNN Inception V3 classifier, which obtained 97% recall, precision and F1-Score, using visual programming software.  

    This study contributed to the advancement of knowledge in health-related technologies by exploring new methods to aid in the diagnosis of laryngeal cancer. The artificial intelligence solution based on CNN with the proposed visual programming reached competitive results, allowing the detection of laryngeal lesions in all images presented to the model, with the advantage of being implemented in an easy-to-use tool, with a lower cost of computational resources and because it does not require advanced knowledge in programming machine learning models, compared to the other methods already explored. 

6
  • Hélder Line Oliveira
  • STUDY AND SIMULATION OF WIRELESS ENERGY TRANSFER THROUGH NEAR FIELD AND INDUCTIVE COUPLING FOR RECHARGING BATTERIES APPLIED TO IMPLANTABLE MEDICAL DEVICES

  • Líder : GERARDO ANTONIO IDROBO PIZO
  • MIEMBROS DE LA BANCA :
  • GERARDO ANTONIO IDROBO PIZO
  • LEANDRO XAVIER CARDOSO
  • WELLINGTON AVELINO DO AMARAL
  • RAFAEL FONTES SOUTO
  • Data: 25-may-2023


  • Resumen Espectáculo
  • The use of implanted medical devices has been a reality since the 1950s and continues to be considered a necessary solution for monitoring biological functions or for the treatment of situations that require medication application or organ stimulation. However, the limitation of the energy supply, primarily the battery, makes the use of these devices uncomfortable for patients due to the need for surgical procedures for their maintenance. In the presented context, wireless energy transfer technology emerges as a proposal to minimize this inconvenience, providing greater reliability in the use of these devices. The work simulated this procedure, seeking to identify the necessary parameters for its application in implantable devices and confronting them with safety standards proposed by ANVISA and ABNT. To do so, a quantitative approach was used, researching information that corroborates the use of the inductive coupling method in battery charging with the application of the specifications found in the simulation performed in Maxwell HFSS®. The tests conducted demonstrated that the temperatures generated on the skin and the device remained around 20ºC, a value below the maximum acceptable for the human body, and that the specific absorption rate was 0.0851W/kg below the conventionally set minimum. Furthermore, the obtained electric voltage and current were 3.25V and 3.20mA, respectively, and within the desired limits. Based on the adopted parameters and current standards, the results obtained indicate that this technology is a viable option for recharging batteries in implantable medical devices, also considering safety in its application to patients.

7
  • ANNE KAROLINE FEITOZA MENDONÇA DE FREITAS
  • Dynamical description of Pandemic COVID-19 in Brazil by Fractional Epidemiological Models

  • Líder : RONNI GERALDO GOMES DE AMORIM
  • MIEMBROS DE LA BANCA :
  • RONNI GERALDO GOMES DE AMORIM
  • GERARDO ANTONIO IDROBO PIZO
  • RODRIGO ANDRES MIRANDA CERDA
  • GESIEL GOMES SILVA
  • Data: 02-jun-2023


  • Resumen Espectáculo
  • In this work it was studied the dynamics of COVID-19 spreading in Brazil using fractional epidemiological models.  It analyzed the fractional version of SIR, Lotka-Volterra and logistic models.  The computational approach was implemented by the fractional Adams-Brashforth method. As  a result, it is possible to conclude that the fractional derivative improves the description of reality by mathematical modeling. 

8
  • Giselle de Oliveira Lima
  • EVALUATION OF DEVELOPMENT AND APPLICATION OF TECHNOLOGIES FOR HEALTH: STUDY CASE AND AN INTEGRATIVE REVIEW

  • Líder : SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • ADSON FERREIRA DA ROCHA
  • ALDIRA GUIMARAES DUARTE DOMINGUEZ
  • ALLISSON LOPES DE OLIVEIRA
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • Data: 16-jun-2023


  • Resumen Espectáculo
  • Introduction: Information is a fundamental tool nowadays, being essential for control, data rationalization, management quality, resources, among others. A short-term view is noted, with regard to the internal culture of health units, generating high costs in investments in information technology. Although Brazil has a wide experience, there is a lack of monitoring of the advancement of technologies, generating concern in devising strategies for use in the Unified Health System (SUS), especially in the area of immunization, which leads to a lack of motivation by those who use it. The application of these technologies will allow overcoming the technological gap, enabling the creation of a control and monitoring process. The cards proposed with Radio Frequency Identification Technology (RFID) are intended to provide better control of the reception process to the user, increasing the accuracy of information, and people control management, improving the quality of life of both the user and the of health professionals. Objective: To approach the manufacturing project of the Totem and SUS+ Card and ImunaSUS module as a resource to be thought of as a technology that facilitates the lives of patients, analyzing, from the Integrative Literature Review, the scientific evidence, on which tools are being applied to digital immunization control. Methodology: The present research will express two moments to be discussed: the first will present a case study on the proposal of the University of Brasília for the confection of a system that uses a patient data storage card with RFID technology, the Totem SUS+ and ImunaSUS module, describing the adopted methods and necessary materials to produce them. Next, it will show the steps taken to produce an integrative review of textual literature in which literature that talks about similar technologies will be analyzed. Results: The implementation of these equipment, SUS+ and ImunaSUS, which is based on the association of a card with a reader and recorder and a self-service totem, will enable agility in the assistance of the SUS user, ease in the acquisition, organization and retrieval of information in the health databases. With regard to technologies that are being used to control information related to vaccines, little development of technologies can be seen in the consulted literature, not only in Brazil, but in the world. Conclusion: The proposal developed by UnB is relevant, innovative, similar to the articles found, suggesting tests with products in operation in health environments and in the SUS.

9
  • JACKSON PAZ BIZERRA DE SOUZA
  • Analysis of clinical electrode applied in ablation procedure with geometry optimization via artificial intelligence

  • Líder : SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • ADSON FERREIRA DA ROCHA
  • ALLISSON LOPES DE OLIVEIRA
  • RONNI GERALDO GOMES DE AMORIM
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • Data: 20-jun-2023


  • Resumen Espectáculo
  • The process of ablation through electrodes aims to necrotize damaged or abnormal tissues in the body. This technique is commonly used to treat conditions such as tumors and cardiac arrhythmias. During the procedure, an electrode is inserted into the area to be treated, and radiofrequency energy is applied, heating and destroying the tissue. Recent advancements include the use of artificial intelligence to guide real-time ablation, the development of more precise ablation devices, and the identification of new therapeutic targets for tumor treatment. Although there has been extensive study on the application of various electrode geometries, there is still a gap in the correlation between electrode geometry and temperature outcomes in tissue ablation. To fill this gap and improve the ablation process, a methodology for optimizing clinical electrodes used in ablation procedures is proposed. For this purpose, the artificial intelligence clustering algorithm K-Means is utilized. Using finite element methodology and parameters from the bio-heat transfer equation (Pennes equation), a needle electrode for tumor ablation was implemented in COMSOL software. The correlation between electrode geometric data, time ranges, voltage ranges, and the resulting temperature range between 45°C to 55°C (318.15K to 328.15K) was evaluated aiming to enhance the efficiency and precision of the ablation technique. The obtained results are compared and analyzed using statistical techniques and machine learning clustering (scatter plots), allowing for the identification of optimal electrode configurations. Preliminary results indicate a convergence between the methodology and the presented results, as denoted in graphs, focusing on the concentration of the target temperature over time and application. The dissertation contributes to the advancement of the field of electrophysiological ablation, bringing an innovative and promising approach to the optimization of clinical electrodes used in these procedures.

10
  • Daniel Leal Fagundes
  • Mitochondrial Bioenergetics and Behavioral Effects of Intermittent Fasting in Zebrafish

  • Líder : JAIR TRAPE GOULART
  • MIEMBROS DE LA BANCA :
  • JAIR TRAPE GOULART
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • VERA REGINA FERNANDES DA SILVA MARAES
  • MAURO EUGÊNIO MEDINA NUNES
  • Data: 21-jun-2023


  • Resumen Espectáculo
  • The current lifestyle and dietary habits contribute to obesity, type 2 diabetes, and cardiovascular diseases. Various approaches are being used globally to combat these diseases, including drugs, alternative diets, and intermittent fasting (IF). IF has gained popularity and has shown positive outcomes like weight loss and reduced insulin resistance. We conducted a study using a zebrafish model to investigate the impact of IF on mitochondrial function. The fish were divided into three groups: a standard diet (SD) group with unrestricted eating, a group with 24 hours of fasting alternated with 24 hours of eating (IF24), and a group with 48 hours of fasting alternated with 24 hours of eating (IF48). After six weeks, we measured the animals' body weight and performed morphometric measurements and organ dissection for high-resolution respirometry to measure oxygen consumption in cardiac and brain tissues. Our results revealed that IF influenced the animals' morphometric parameters. Mitochondrial dysfunction was not observed in the hearts and brains of animals in the IF24 group, but we observed decreased cardiac mitochondrial metabolism efficiency in the IF48 group. Overall, our findings support the effectiveness of IF in combating obesity, particularly when using a protocol of alternating 24 hours of fasting with 24 hours of eating ad libitum.

11
  • Carolina Ramos dos Santos
  • Extraction, characterization and evaluation of in vitro cell regeneration of crude seed extract of Bixa orellana L

  • Líder : MARCELLA LEMOS BRETTAS CARNEIRO
  • MIEMBROS DE LA BANCA :
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • JAIR TRAPE GOULART
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • MARCUS VINICIUS LIA FOOK
  • Data: 21-jun-2023


  • Resumen Espectáculo
  • Diabetic Foot Ulcer (DFU) represents a high risk factor for amputation and is among the leading causes of hospitalization in diabetics, culminating in high cost to the Brazilian Unified Health System (SUS). Thus, this disease represents a serious public health problem, raising the need for public investment in the development of research aimed at finding curative therapeutic strategies. Our group developed a therapeutic system for tissue neoformation which was named Rapha® and consists of the concomitant use of phototherapy and curcumin-based natural latex biomembrane (BLN). The results were very promising, since healing with this system was superior to conventional SUS treatment. However, this dressing did not show antimicrobial effect. Considering that bacterial infections are related to wound worsening and increased chances of amputations, it is proposed here, the prospection of medicinal extract with antioxidant and antimicrobial properties. In this work, crude extract of "Urucum" (Bixa orellana) seeds will be used, obtained commercially through donation by New Max® and Purifarma® or through solvent extraction. The annatto seeds used to obtain the extract were obtained from an extracting quilombola community. The extraction of the compounds from the annatto seeds was performed using the solvents Hexane, Ethyl Acetate and Ethanol at 50°C and 130°C to define the most efficient extraction protocol. For evaluation of the chemical composition of the extract and validation of the extraction method, analyses were performed using High Performance Liquid Chromatography (HPLC) and Spectrophotometry. According to the results, the extraction performed with Ethanol, at a temperature of 50°C, was more efficient, since it presented higher contents of carotenoids such as bixin and norbixin. In the analysis by CLAE, the sample presented 57.5% of bixin. The analysis performed by spectrophotometry, at 487 nm, was performed to calculate the % peak area, and thus calculated the content of bixin and norbixin. For the sample extracted by Ethanol at 50°C the content of 0.27% bixin and 0.30% norbixin content was demonstrated. The biocompatibility of annatto extracts was evaluated, using the MTT reagent ([3-(4,5-dimethylthiazol-2yl)-2,5-diphenyl tetrazolium bromide), in fibroblast and keratinocyte cultures at doses of 80 to 2.5 µg/mL of annatto extract diluted in DMSO or Ethanol for 24, 48 and 72 h. The results showed that the concentrations of 2.5, 5 and 10µg/mL were viable, showing cell viability greater than 80% in both cell lines. In addition, the therapeutic potential of the extracts on regeneration was investigated, by means of cell migration assay (Scratch assay) where, in 72 hours, the treatment at the 10µg/mL concentration decreased around 90% of the groove made in fibroblast cells. The antioxidant potential of annatto extracts was determined by means of the Reactive Oxygen Species (ROS) production assay using the DPPH (2,2 Diphenyl 1 picrylhydrazylo) marker. DPPH was left in contact with the sample of annatto seed extract for 30 minutes at concentrations of 10 and 100µg/mL. Although there was no significant difference between the concentrations tested, both showed a percentage of potential antioxidant activity around 50%. As a positive control, ascorbic acid was used in the same concentrations, which presented a percentage above 90% for both concentrations.

12
  • FERNANDA MAYUMI GUEDES FUKUOKA
  • Development and analysis of a RAPHA device with blue LED associated biomembrane of latex derived from Hevea brasiliensis

  • Líder : SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • ALDIRA GUIMARAES DUARTE DOMINGUEZ
  • RONNI GERALDO GOMES DE AMORIM
  • LUCIANA ROBERTA TENORIO PEIXOTO
  • Data: 26-jun-2023


  • Resumen Espectáculo
  • Diabetes Mellitus (DM) is a chronic health problem that interferes in the healing process. This interference generates the worsening of ulcers, especially in the lower limbs, causing infections and possibly leading to limb amputation. There are technologies that help in the healing process, such as photobiostimulation. The development of new technologies is important to improve the quality of life of patients. In this work we developed a photobiostimulation equipment with blue LEDs, based on the RAPHA equipment with red LEDs. The blue color was chosen for its bactericidal characteristics. The equipment was developed at the University of Brasilia (UnB) and the constructive parameters were analyzed in the laboratory. The characterization of the equipment and its interaction with the latex blade was done, to determine the wavelength and power. The equipment with blue LED presented a higher power than the red one.

13
  • Jessica Sousa Oliveira
  • Web Scraping in the extraction and systematic combination of contents: an auxiliary tool in Research, Development and Innovation (RD&I) processes.

  • Líder : MARIO FABRICIO FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • MARIO FABRICIO FLEURY ROSA
  • ADSON FERREIRA DA ROCHA
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • Paulo Roberto dos Santos
  • Data: 29-jun-2023


  • Resumen Espectáculo
  • Introduction: The Legal Framework for Science, Technology, and Innovation aims to contribute to the sustainable development of Brazil by coordinating and allocating resources to ensure scientific, technological, industrial, and commercial products or studies in the country. For public institutions of higher education to benefit from the rights guaranteed by this framework, it is necessary for them to demonstrate and make public their contributions to society, including the translation of scientific development for social benefit. In this regard, there is a demand for a systematic search that emphasizes the scientific and technological initiatives developed by these institutions, which subsequently result in technical and technological products beyond publications, but mainly patents and/or technology transfer. Objectives: Therefore, the development of an application that automates this process emerges as a viable technological solution. This work aims to deliver a data mining tool based on web scraping that performs systematic and practical data collection by extracting and combining web content, seeking authors of projects developed at the University of Brasília who have subsequently translated their results into patents and/or technology transfers. Methodology: To this end, an applied approach was used, aiming to solve a specific problem involving local interests. Based on literature reviews and similar research products, it is expected that the scraper will facilitate the extraction of information about research projects deposited in a specific online repository and, by combining the obtained data with those mined from a database of patents and/or technology transfers, demonstrate the return of studies, productions, and research to the country's economic and social development. Results: As a result of executing the tool on the INPI repository, 783 names of patent inventors were obtained, and the application on the UnB repository returned 53,704 authors of projects developed at the university. After comparing the generated tables, 3,244 records referring to authors who filed a patent were considered, which assists in the analysis of research that became innovations. Conclusion: The outcome indicates that the developed tool is functional for its intended purpose, considering that the cross-referencing of information is entirely feasible and can be incorporated into analyses and/or process improvements that maximize the innovation indicators generated by universities, mitigating the Legal Framework for Science, Technology, and Innovation.

14
  • William Carneiro de Mendonça
  • FUEL CELLS AS A SOURCE OF ENERGY FOR IMPLANTABLE BIOELECTRONIC DEVICES: A SYSTEMATIC REVIEW

  • Líder : RONNI GERALDO GOMES DE AMORIM
  • MIEMBROS DE LA BANCA :
  • RONNI GERALDO GOMES DE AMORIM
  • GERARDO ANTONIO IDROBO PIZO
  • MARCELO BENTO DA SILVA
  • ROBERTO KENNEDY FERREIRA DA SILVA DE QUEIROZ
  • Data: 30-jun-2023


  • Resumen Espectáculo
  • Introduction: FUEL CELLS AS ENERGY SOURCES IN IMPLANTABLE MEDICAL DEVICES: A SYSTEMATIC REVIEW consists of a bibliographic review on fuel cells and their degree of technological development and possible applications in biomedical engineering regarding active implantable medical devices. The fuel cell is an electrochemical device that uses hydrogen gas and oxygen gas to transform chemical energy into electrical energy, silently and without environmental hazards; It is also the objective of this work to highlight the importance of this new technology in the global energy scenario, with emphasis on Brazil, since, as it is a clean energy source, with zero or almost zero environmental impact, it is a strong candidate for replacing the current fossil energy matrix. Objective: This work explains, in general lines, the origin, functioning of this technology, as well as its advantages, limitations and applications with emphasis on biomedical engineering with regard to self-sustaining implantable bioelectronic devices. Methodology: Initially exploratory bibliographic research followed by a systematic review using the PICO strategy, an acronym for Patient, Intervention, Comparison and “Outcomes” (outcome). The research was carried out on a digital journal platform: a) Portal de Periódicos da Capes, b) Pubmed, c) Embase, d) IEEE Xplore, e) Scopus, f) Science Direct. The timespan of the research was between 2021 and 2022. For the inclusion and exclusion criteria of articles, the PRISMA-2015 protocol was used. Conclusion: Research into the development of fuel cells to power Active Implantable Medical Devices is on the rise, but in the development phase. There are already in vitro and in vivo tests. However, many barriers must be overcome for the use of these electrochemical devices in DMIA. The availability of this technology will reduce surgical risks for patients, decrease financial expenses with: medical staff, surgical centers, in addition to eliminating hospitalization time in ICUs.

15
  • INAE RODRIGUES DAMACENO SILVA
  • Chronic effects from the proprioceptive neuromuscular facilitation on reflex responses and V-wave of the soleus muscle

  • Líder : RINALDO ANDRE MEZZARANE
  • MIEMBROS DE LA BANCA :
  • RINALDO ANDRE MEZZARANE
  • JAKE CARVALHO DO CARMO
  • LUCIANA HAGSTROM BEX
  • FERNANDO HENRIQUE MAGALHÃES
  • Data: 07-jul-2023


  • Resumen Espectáculo
  • Increased flexibility can be associated with improved life quality (relief of pain, increased motor performance, etc). Among the techniques used to improve flexibility, there are the static stretch, dynamic stretch, proprioceptive neuromuscular facilitation (PNF), etc. Although the PNF is one of the most effective, little is known about the mechanisms responsible for the effects on flexibility. Despite the assumption of neurophysiological mechanisms mediating the effects, such as those involved in the modulation of the excitability of the stretch reflex pathway reflex (e.g., presynaptic inhibition), the studies on this topic are scarce. The purpose of the present study is to investigate the neurophysiological mechanisms involved in the changes in the maximal ankle joint angle after PNF training using the technic of contract-relaxing (CR). Thirteen healthy young volunteers participated in the present study. The intervention consisted of 16 sessions (4 weeks, 4 times per week). Each session consisted of a passive stretch for 10 s followed by a voluntary contraction for 6 s at 60% the maximal voluntary contraction (MVC), and a second passive stretch of 10 s. The maximal joint amplitude showed significant differences with the PNF training. The additional variables that showed statistical significance were Mmax, the torque related to Mmax and the RMS value from the EMG during the CVM. It is suggested that the mechanisms involved in the improvement of ankle joint flexibility are peripheral, i.e., are not related to central nervous system adaptations.

16
  • ANA KAROLINE ALMEIDA DA SILVA
  • Ethical and regulatory in the use of organs-on-a-chip platforms for diabetic foot wound studies: perspectives from biomedical engineering and translational research in Brazil

  • Líder : SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • GLECIA VIRGOLINO DA SILVA LUZ
  • CICILIA RAQUEL MAIA LEITE
  • JOSÉ CARLOS TATMATSU ROCHA
  • Data: 18-ago-2023


  • Resumen Espectáculo
  • The research and development of Organ-on-a-chip (OoC) devices have been garnering increasing attention due to their revolutionary potential in the field of medicine and biomedical research. These miniature devices are designed to replicate the structure and function of human organs with greater precision than traditional cell cultures or animal models. They offer highthroughput screening, accelerating the discovery of new drugs and identifying potential side effects or interactions between different organs with greater accuracy. Additionally, they have the potential to reduce the need for animal models in toxicological testing. One of the promising applications for OoCs is the study of Diabetes Mellitus (DM) and its complications, such as diabetic foot ulcers (DFU). DFU is a severe comorbidity associated with high rates of morbidity, mortality, and amputations, posing a major challenge to global healthcare systems. The use of this technology in this context aims to provide crucial insights into the pathophysiology and enable more effective therapeutic approaches, including personalized medicine. This dissertation aims to analyze the ethical and regulatory aspects of implementing OoCs in Translational Health Research (THR) in Brazil, with a focus on the role of Biomedical Engineering in developing this technology. The work is divided into seven chapters, covering various elements related to the devices and their application in the context of diabetic foot ulcers, including two systematic reviews. The first article focuses on the use of microphysiological systems in the DFU scenario, exploring their applicability in understanding functional changes in the body and the underlying disease mechanisms. The second study sought evidence on a potential healing biocompound, urucum, for the treatment of chronic wounds. Additionally, national and international legislations and regulations governing the use of OoCs were analyzed. Furthermore, a scientific marketing plan was developed to disseminate knowledge about these platforms and their significance in scientific research and innovation. This comprehensive methodology allowed exploring different aspects of OoCs and their relevance to DFU studies, as well as promoting the advancement of this technology in national THR. This dissertation represents a significant milestone in the field of legislation and regulation of devices in Brazil. It aims to provide a solid foundation that can be used in shaping public policies and regulatory guidelines in the country. The importance of interdisciplinarity between Biomedical Engineering and the healthcare sector is emphasized, propelling Brazil to the forefront of biomedical research and ensuring that the use of OoCs benefits not only science but society as a whole. This work serves as a call to action for regulatory authorities, researchers, and healthcare professionals to join forces in creating an agile and efficient regulatory environment that encourages the safe and ethical advancement of OoCs, paving the way for a new era of innovative discoveries and treatments with the potential to positively transform the health and quality of life of millions of Brazilians.

17
  • Denilson Nogueira dos Santos Paixão
  • Analysis of tumor Dynamics by fractional differential equations

  • Líder : RONNI GERALDO GOMES DE AMORIM
  • MIEMBROS DE LA BANCA :
  • RONNI GERALDO GOMES DE AMORIM
  • LEANDRO XAVIER CARDOSO
  • RODRIGO ANDRES MIRANDA CERDA
  • RENDISLEY ARISTÓTELES DOS SANTOS PAIVA
  • Data: 15-sep-2023


  • Resumen Espectáculo
  • In this work, we address one of the major health problems that plagues humanity: cancer. Currently, there are several studies about the treatment and combat of this disease. In this sense, this work analyzed some mathematical models of the evolution of breast cancer by applying fractional calculus, according to Caputo's fractional derivative approach. For this, an analysis of the fractional versions of the following mathematical models was carried out: the Malthus model; Logistics equation; and Gompertz equation. Then, we solved the fractional differential equations numerically using the fractional Adams-Bashfort method. As a result, we conclude that methods based on fractional calculus describe tumor growth more satisfactorily than methods based on integer order calculus. And yet, among the three fractional models analyzed, the fractional logistic model was the one that presented the best correlation with the real data, that is, when compared to experimental data, it presented the lowest mean squared error.

18
  • Allan Paulo de Souza
  • Characterization of Electroencephalography Patterns Using Machine Learning in the Identification of Sleep-Associated Pathologies

  • Líder : CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • ADSON FERREIRA DA ROCHA
  • NILTON CORREIA DA SILVA
  • LUCIANO MANHÃES DE ANDRADE FILHO
  • Data: 28-sep-2023


  • Resumen Espectáculo
  • Sleep consists of a series of fundamental stages for the quality of life in several species. Its importance is comparable to food and physical activity. However, their understanding still has gaps that need to be studied.

    In this sense, two technologies can contribute, in particular, to the understanding of sleep. The first concerns the electroencephalography (EEG), which is relevant in several neuroscience studies, including sleep. The other technology is machine learning, which has been gaining ground as a tool for identifying patterns that aid in diagnosis; but there are still few works that use these techniques to identify pathologies associated with sleep.

    Therefore, this work proposes to compare different machine learning approaches from EEG data in order to automate analyzes to aid in the diagnosis of sleep-associated pathologies. For this, the use of support vector machines (SVM) and random forests (Random Forest) with different hyperparameters from characteristics obtained by calculating energies in signal frequency bands were compared.

    In experiments with SVM, approaches with grain variation were tested; with Random Forests, maximum depth. Furthermore, these results were compared with those obtained using convolutional neural networks (CNN), defining characteristics dynamically through the layers of this network. In these CNN architectures, distinct convolutional and dense layers were used with changing hyperparameters.

    To carry out this work, polysomnography data available on the Physionet portal was used. From these raw data, there was feature selection and, with the exception of experiments that used CNN, feature extraction using Fourier transform, selecting different derivations (electrodes).

    In the pilot study, algorithms were implemented to classify the EEG signal of nocturnal frontal lobe epilepsy. The results obtained with SVM and Random Forest suggest that these classifiers can identify relevant patterns from EEG signals: the highest accuracy (mean of 5 folds) was 60% in both approaches. In addition, comparisons of the work with these approaches showed that the spatial distribution of information is spatially homogeneous. On the other hand, it was not possible to distinguish the EEG signals of participants with epilepsy, even using different architectures, varying the number of layers, activation functions, number of neurons and dropout.

    Then, new experiments were implemented using the results of preliminary studies as a guideline. For this, new classifiers were trained with data that went through other stages of preparation from signal windowing and signal filtering. As a result, there was an improvement in the performance metrics of the classifiers.

19
  • Josue Nascimento da Silva
  • Automatic Generation of X-ray Images Diagnostic Reports, with Attention-Based Explainability Applied to a Recurrent Neural Network

  • Líder : CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • NILTON CORREIA DA SILVA
  • FABIANO ARAUJO SOARES
  • ANDRÉ COUTINHO CASTILLA
  • Data: 29-sep-2023


  • Resumen Espectáculo
  • Specific problems related to the analysis of radiological exams have been widely documented for at least 50 years. Among the main circumstances that lead to diagnostic errors, evaluations conducted by early-career physicians, inadequate communication among team members, night shifts, shift changes, and faulty reasoning are noteworthy.

    In this regard, the use of artificial intelligence as a tool for decision-making and diagnosis has the potential to assist healthcare professionals in achieving greater precision and sensitivity in their analyses, thereby improving patient treatment. Therefore, the objective of this study is to develop an encoder-decoder architecture for an artificial intelligence model capable of automatically generating medical reports with specific information extracted from image exams. The idea is that this information in the images reflects the aspects guiding the decisions and analyses indicated in the report text, representing a contribution compared to the predominant approaches in the literature, which often focus solely on the text itself.

    To accomplish this objective, X-ray images along with their respective reports were used. An encoder network based on the Densenet121 architecture was developed to extract features from the exams, which are then translated by a decoder based on transformers, allowing for the learning of semantic relationships between words, along with the long short-term memory (LSTM) technique for report generation.

    To relate regions of the images to the generated words, the spatial attention technique was applied, capturing the most relevant regions for producing specific words by the model. This process was applied to five pathologies: lung hypoinflation, lung hyperdistention, cardiomegaly, aorta tortuous, and spine degenerative, resulting in five encoder-decoder networks. During training, F1-score values of 76\% and area under the curve (AUC) of 80\% were achieved for the encoder. The encoders were evaluated using cross-validation to assess their generalization capacity with respect to the data used. As for the decoder, in report production, it was evaluated using the recall-oriented understudy for gisting evaluation (ROUGE) metric, obtaining average values of 0.32.

    In conclusion, the proposed architecture is capable of generating reports and annotations in the image exams, serving as support for medical decision-making.

20
  • WESLAINE MACEDO GUIMARAES DOS SANTOS
  • "Integration between the Legal Framework for Innovation and Biomedical Engineering: the case of the University of Brasília"

  • Líder : MARIO FABRICIO FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • MARIO FABRICIO FLEURY ROSA
  • ADSON FERREIRA DA ROCHA
  • GLECIA VIRGOLINO DA SILVA LUZ
  • JOSÉ CARLOS TATMATSU ROCHA
  • Data: 23-nov-2023


  • Resumen Espectáculo
  • Introduction: The present work aims to analyze the impact of the Innovation Regulatory Framework on the Research, Development and Innovation (RD&I) process in Health within the scope of the University of Brasília (UnB), in the context of research projects managed by the Development Center Technological Innovation (NIT) at UnB in partnership with the Deanship of Research and Innovation (DPI), for the translation of technologies to the health area linked to the area of Biomedical Engineering. Objectives: The objectives of this work are: i. analyze the interface between the Legal Framework for Innovation and researchers linked to the public university; ii. Analyze the contributory mechanisms between the Legal Innovation Framework and the Technological Innovation Centers of public universities; iii. categorize research projects institutionalized at the University of Brasília through the Chamber of Projects, Agreements, Contracts and Related Instruments (Capro) of the Deanship of Research and Innovation (DPI) in the area of Biomedical Engineering; iv. correlate research projects with patent filings and/or technology transfers. Methodology: The descriptive, documentary and experimental method will be used, through the qualitative and quantitative analysis of processes, policies, databases and existing documents. In the background, the types of contracts and partnerships signed in the area of biomedical engineering and contemplated in the years 2011 to 2021 will be mapped with a view to identifying probable gaps within the scope of the Legal Framework and their consequences in the UnB RD&I process. Results: Numerous research projects, within the scope of UnB, managed to achieve their purpose, generating royalties and technology transfer, thus fulfilling the translation of research. However, very few projects are linked to the health area in Biomedical Engineering. Conclusion: Based on the data presented, it is possible to provisionally conclude that the University of Brasília has a large scientific and technological production, with emphasis on other areas of research besides health. It is possible to conclude that, although the University of Brasília operates through innovation and technological transfer management bodies, such as ACT, CITT, CDT and DPI, generating a notable number of research projects, technological transfers and deposits of patents, projects aimed at the health sector are still substantially reduced, particularly in the context of Biomedical Engineering.

21
  • JESSICA AGUIAR CARNEIRO
  • "ANEMIA AND LONG PERIODS OF ICU ADMISSION"

  • Líder : GLECIA VIRGOLINO DA SILVA LUZ
  • MIEMBROS DE LA BANCA :
  • GLECIA VIRGOLINO DA SILVA LUZ
  • ALDIRA GUIMARAES DUARTE DOMINGUEZ
  • SOLANGE BARALDI
  • THAIS MARTINS GOMES DE OLIVEIRA
  • Data: 25-nov-2023


  • Resumen Espectáculo
  • ABSTRACT: Anemia, despite being one of the most common diseases in the world, is of great importance because, in addition to affecting about a quarter of the world's population, it can seriously impair the clinical status of critically ill patients. There are several causes of the disease in the ICU, such as successive phlebotomies, hemolysis, irregularity in iron metabolism, among others. However, one of the sources of anemia still little studied in academia is the loss of blood volume due to daily collection for laboratory tests, which, in the vast majority, does not help close the expected diagnosis. The study aims to analyze the possible factors involved in the clinical manifestation of anemia in patients hospitalized for long periods in the Intensive Care Unit. By describing the demographic and clinical profile of critically ill patients diagnosed with anemia, analyzing whether routine health behaviors influence the manifestation of this disease, and reporting the characteristics and risk factors that predispose to this disease. The development of the research will take place at the Intensive Care Unit of the Hospital Regional de Santa Maria (HRSM) and at the Hospital Regional do Gama (HRG) in the Federal District (DF), using a quantitative approach and observational and longitudinal design. Including Men and women, admitted to the ICU, aged between 18 and 59 years of age; hospitalization period equal to or greater than 7 days and excluding Diagnosis of anemia at the time of admission to the ICU; outcome (considering discharge or death) of the patient within a period of less than 7 days. All information will be collected, expressed, managed, and disseminated by the REDCap software.

22
  • GEOVANNI OLIVEIRA DE JESUS
  • EXPLAINABLE ARTIFICIAL INTELLIGENCE MODEL FOR MAMMOGRAM BREAST CANCER CLASSIFIERS

  • Líder : CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • NILTON CORREIA DA SILVA
  • FABIANO ARAUJO SOARES
  • LUCIANO MANHÃES DE ANDRADE FILHO
  • Data: 30-nov-2023


  • Resumen Espectáculo
  • Breast cancer constitutes the highest incidence rate of cancer among women worldwide. As with other types of cancer, the faster it is discovered, the less invasive the treatment, and the greater the chances of the patient surviving. Having a tool that helps the analysis of mammograms to discover breast lesions and their classification is a way to help prevent cancer. With the development of technology, Artificial Intelligence has also grown, as well as its ramifications. Machine Learning and Deep Learning have developed and achieved good performance in tasks such as image classification and object detection. The mammogram analysis is an image classification task, and having a tool that helps in this task, which can be more than a black box classification in which it is not known how the response is obtained if the Machine Learning model were capable of explaining how it has achieved at the final classification, it would be possible to better direct physicians to use this model in their day-to-day analysis of mammograms.

    Currently, some solutions address the classification of breast lesions, however, they use traditional Machine Learning algorithms, which do not have a performance as high as Deep Learning algorithms. In addition, they are considered black box solutions, which means that an answer is given from an input image, and it is not possible to know which factors influenced the final result of the classification. An Artificial Intelligence model that has its results explainable, that is, it is possible to understand what factors led it to reach the final classification result, gives more reliability to the users who use it. With this prediction explanation, it is possible to comply with laws such as the General Personal Data Protection Act (LGPD) and even the General Data Protection Regulation, the acronym in English GDPR, which is the European data protection law. An explainable model can fit into laws and make the solution be applied in a real domain. The diagnosis of cancer is a sensitive moment, and because of this, knowing how a Deep Learning or Machine Learning model reached a certain result can better direct doctors to investigate cases in a more targeted way, thus giving more emphasis to some characteristics of the image, in addition to generating more reliability in model prediction results.

    The use of Deep Learning for image classification tasks has obtained surprising results, which match and in some cases even exceed human capacity, which is why this approach will be discussed in this dissertation. Combining a powerful classification tool with techniques that allow leaving the created models with explainable predictions, thus making a breast lesion classification tool with high-reliability potential for its end users, the medical specialists. To achieve these goals, Deep Learning architectures such as VGG16 and explainability techniques of trained models such as LIME, which is a well-performing and simple-to-use explainability framework, are investigated.

    This dissertation aims to develop a Deep Learning model that uses Explainable Artificial Intelligence (XAI) techniques, that is, it has explainable predictions that classify breast lesions and identify the important characteristics that led the model to achieve such a result. result.

    After model training using VGG16 architecture, the analyzed metrics were accuracy, specificity, and sensitivity, the results obtained were respectively 68 %, 77 %, and 65 %. Higher results were found in the literature, but these are not reproducible results. In many cases, the databases are private to hospitals where the team surveyed mammograms over the last 20 years, created the dataset, and performed the tests. There was a master's thesis done in 2020 by Adam J. with a similar approach and which obtained results close to those presented in this dissertation. The author reported an accuracy result of 67 %.

23
  • PAULA LAUANE ARAUJO
  • "Technological development and characterization of a wound dressing based on Natural Latex (Hevea brasiliensis) containing bixin extract (Bixa orellana L.)" 


  • Líder : MARCELLA LEMOS BRETTAS CARNEIRO
  • MIEMBROS DE LA BANCA :
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • GLECIA VIRGOLINO DA SILVA LUZ
  • JOSÉ CARLOS TATMATSU ROCHA
  • MARCUS VINICIUS LIA FOOK
  • Data: 07-dic-2023


  • Resumen Espectáculo
  • Úlcera de Pé Diabético (UPD) está entre as principais causas de hospitalização em diabéticos em decorrência de complicações associadas a altas taxas de amputação. Além disso, UPD representa alto custo para o Sistema Único de Saúde (SUS), representando um grave problema de saúde público. Assim, destaca-se a necessidade de investimento público para o desenvolvimento de pesquisas voltadas para a busca de estratégias terapêuticas curativas. Nosso grupo desenvolveu um sistema terapêutico para neoformação tecidual, denominado Rapha® o qual consiste no uso concomitante de fototerapia e curativo baseado em biomembrana de látex natural (BLN), oriundo de Hevea brasiliensis, com curcumina, um biocomposto com atividade antiinflamatória. Os resultados clínicos foram muito promissores, visto que a cicatrização foi cerca de 60% superior em relação ao tratamento convencional do SUS. Todavia, este curativo não apresentou efeito antimicrobiano. Considerando que infecções bacterianas estão relacionadas com o agravamento da ferida e aumento das chances de amputações, neste trabalho será desenvolvido um curativo baseado em biomembrana de Hevea brasiliensis (BLN) contendo biativos do urucum (Bixa orellana L), a fim de avaliar sua segurança e seu potencial terapêutico relacionado à cicatrização bem como seus efeitos antimicrobianos em ratos com infecção induzida. O bioativo utilizado será a bioxina, que consiste em um extrato fitocomplexo, extraído da semente do urucum e com alta biodisponibilidade, padronizado em geranilgeraniol e tocotrienóis. A bioxina foi obtida de forma comercial por meio de doação pela empresa Purifarma®. Para a produção das biomembranas, látex natural e bioxina foram adicionados em um béquer e homogeneizados com auxílio de agitador magnético, Em seguida, adicionou-se água destilada, em temperatura ambiente e novamente homogeneizado em agitador magnético. A solução resultante foi transferida a tubos falcon e centrifugada sob alta rotação durante 10 minutos. Após centrifugação, a solução foi coada e adicionada em placas de petri de 90 mm de diâmetro com auxílio de pipetador automático. As placas foram acondicionadas em ambiente ventilado para repouso durante 48 horas e seguiram para estufa de secagem durante 1 hora. Após vulcanização, as biomembranas foram retiradas das placas de petri com auxílio de uma pinça e embrulhadas em papel manteiga e, em seguida, acondicionadas em papel grau cirúrgico. As biomembranas produzidas passaram por testes para controle de qualidade.

24
  • Kassia Costa Fernandes
  • Customized orthopedic insoles for physical activity practitioners: a review of the literature


  • Líder : VERA REGINA FERNANDES DA SILVA MARAES
  • MIEMBROS DE LA BANCA :
  • JOSICÉLIA ESTRELA TUY BATISTA
  • JULIANA DE FARIA FRACON E ROMAO
  • MARILIA MIRANDA FORTE GOMES
  • VERA REGINA FERNANDES DA SILVA MARAES
  • Data: 07-dic-2023


  • Resumen Espectáculo
  • Introduction: The use of customized orthopedic insoles has been studied and indicated by health professionals in order to avoid injuries due to postural derangement, anatomical changes, limb dysmetria, pain, among other factors. However, there is still no consensus on the benefits of this intervention for athletes. Objective: To evaluate whether the use of a personalized insole in adult men who run improves the distribution of plantar pressure when compared to the common insole. Method: Systematic review, with search conducted in the following databases: Medline (via Pubmed), Embase, Web of Science, Virtual Health Library, Scopus, SportDiscus and Proquest. Intervention studies that used baropodometry assessment in men who run. Studies with high-performance athletes and people who had some morbidity were excluded from the review. There was no restriction regarding the period of publication and language of origin of the studies. To assess the risk of bias, we used the Robins-I and Rob 2.0 tool from the Cochrane collaboration. Meta-analyses of random effects, using the hedging technique, were conducted to measure the standardized mean difference and respective 95% confidence interval. Results: 1171 studies were detected, of which 13 were part of the systematic review. The sample included 434 male runners, aged 19 to 53 years. The studies were conducted and published between 2004 and 2021. Most of the non-randomised trials were at high risk of bias and the only randomised controlled trial was at high risk of bias. In addition, the meta-analysis indicated that there was no statistically significant difference between the angle of peak dorsoflexion between male runners who used custom insoles when compared to those who did not. Conclusion: The use of insoles did not show significant benefit for athletes. However, few studies were detected, and most of them were at high risk of bias. In this sense, it is recommended that new studies be conducted, with methodological robustness, in order to ratify or reject the hypothesis.

25
  • Cristiano Drumond Ribeiro
  • Artificial Intelligence (AI), Patient safety, Hospital Institutions, Patients, Adverse Events (AE), Quality of healthcare services.

  • Líder : GLECIA VIRGOLINO DA SILVA LUZ
  • MIEMBROS DE LA BANCA :
  • GLECIA VIRGOLINO DA SILVA LUZ
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • RONNI GERALDO GOMES DE AMORIM
  • MARIA LIZ CUNHA DE OLIVEIRA
  • Data: 11-dic-2023


  • Resumen Espectáculo
  • Introduction: Patient safety is an object of great concern in healthcare, as it is directly related to the quality of care. One of the biggest management challenges is developing strategies to reduce the occurrence of errors, which directly impact the quality of care, the individual's quality of life and care costs. The use of technologies such as artificial intelligence has demonstrated good results in solving problems and studies have focused on the analysis and construction of technological tools in the health sector. Objective. Promote the early identification of risk factors for adverse events in hospitalized patients, by facilitating the implementation and use of patient safety protocols, with the use of artificial intelligence. Method. This research was carried out in two parts. The first part refers to the Systematic Literature Review, with the aim of gathering evidence on the use of artificial intelligence in patient safety. The second part consists of the development of software, applied to the Adverse Events Scale Associated with Nursing Practices (EEAAPE), to identify the risk of adverse events occurring in the hospital environment. Results. The search in the databases returned 164 articles, of which, after analysis according to inclusion and exclusion criteria, 14 were included in the review. The data from the articles were extracted and organized in a Microsoft Excel® spreadsheet, and were grouped according to the study design, and the assessment of methodological quality was carried out using tools from the Joanna Briggs Institute (JBI). The software entitled “PrevenSystem - Hospital Incidents” was developed in Java Script language using React Native as a framework, according to the agile development methodology called Scrum. The system's functionality consists of facilitating the identification of risk factors for adverse events, based on the evaluation of items on a standard assessment scale, with the aim of reducing the occurrence of failures and ensuring rapid decision-making. The data entered is processed by algorithms that classify them using the “decision tree” method, so that the frequency of occurrence of a given situation corresponds to a certain probability of failure occurring. This AI application consists of predictive analysis, which is a model that uses machine learning, in an “AI-based prediction model” capable of providing probabilistic predictions of the current presence or future occurrence of a given result, according to the input data. This research includes registering the software with the National Intellectual Property Institute (INPI). Conclusion: The research result demonstrates the relevance of research on the application of AI in the prevention of adverse events in hospitalized patients, with good results in the evaluation of the systems developed, especially in reducing errors and agility in decision-making actions. However, it is identified that there is still a need for new studies in areas not yet addressed by current research. The application developed uses AI to analyze the dimensions of a validated instrument to assess the risk of adverse events occurring, according to the development phases of the agile Scrum methodology, being registered with the INPI, being able to be subjected to testing for validation and wide dissemination for use by health organizations.

26
  • Maria Tereza Dourado Melo
  • Expert System for Identification of Ergonomic Risks in the Surgical Center.

  • Líder : GERARDO ANTONIO IDROBO PIZO
  • MIEMBROS DE LA BANCA :
  • GERARDO ANTONIO IDROBO PIZO
  • MARILIA MIRANDA FORTE GOMES
  • RENATO CORAL SAMPAIO
  • LUCIANA CORREIA ALVES
  • Data: 11-dic-2023


  • Resumen Espectáculo
  • The environment of the Surgical Center, although designed for anestheticsurgical procedures, often does not favor the medical team, leading to injuries and accidents that harm the well-being of healthcare professionals. Osteomuscular symptoms are common due to these unfavorable conditions. In this context, the need arises to create a safe environment for professionals, adhering to ergonomics standards. Technology, especially Artificial Intelligence, emerges as a crucial solution. This study focused on the implementation of an Expert System to identify ergonomic risks in the surgical sector. Data collection was carried out through the analysis of standards and questionnaires. Based on these data, the system was developed using Expert Sinta software and, after testing and validation, proved to be effective in identifying ergonomic risks before the workspace is used by professionals. This allows preventive interventions, improving safety and health in the Surgical Center.

27
  • José Hevenicio do Nascimento Silva
  • Segmentation of Computed Tomography Images Using the U-Net Algorithm: A Case Study in Bladder Images

  • Líder : GERARDO ANTONIO IDROBO PIZO
  • MIEMBROS DE LA BANCA :
  • THAÍNA APARECIDA AZEVEDO TOSTA
  • GERARDO ANTONIO IDROBO PIZO
  • JONES YUDI MORI ALVES DA SILVA
  • RENATO CORAL SAMPAIO
  • Data: 15-dic-2023


  • Resumen Espectáculo
  • The dissertation addresses the influence of bladder filling on the position of the prostate or uterus in pelvic radiotherapy treatments. The reproducibility of radiotherapeutic planning and the dosage of radiation absorbed by the bladder are crucial for the success of the treatment and to minimize toxicity. The project aims to implement the U-Net architecture to segment the bladder in tomographic images, using a set of images from the University Hospital of Brasília (HUB). The U-Net model is expected to significantly improve the efficiency of treatment planning in radiotherapy.

28
  • Cleonice Lisbete Silva Gama
  • Medication distribution profile of the specialized component of pharmaceutical assistance in the Federal District

  • Líder : MARILIA MIRANDA FORTE GOMES
  • MIEMBROS DE LA BANCA :
  • MARILIA MIRANDA FORTE GOMES
  • GLECIA VIRGOLINO DA SILVA LUZ
  • JORGE ANDRES CORMANE ANGARITA
  • THUANY DE ALENCAR E SILVA
  • Data: 21-dic-2023


  • Resumen Espectáculo
  • High-cost drugs within the scope of the Unified Health System represent a complex challenge that involves ethical, political, economic and management issues. Ensuring equitable and efficient access to these medicines is essential for proper patient care. In this context, it is important to analyze the different related aspects, including clinical evaluation, economic impact and public policies, as this analysis can contribute to improving the management and planning of resources, aiming to guarantee the best therapeutic result for patients and financial sustainability of the health system. Objectives: outline different actions for different types of users in order to improve the management of medicines in CEAF DF, based on demographic and epidemiological data, in the years 2012, 2018 and 2021, so that such contributions can serve as a basis for implementing improvements locally and/or or in similar services at the state and municipal level in other locations. Methodology: The present work is characterized as an exploratory study, of a descriptive, observational nature, carried out through bibliographical research with the use of TEMAC, to obtain scientific information of greater relevance, following the bibliometric laws in the main databases, resulting in an integrative approach and analysis of data from a national bank with records of medication use from CEAF.

2022
Disertaciones
1
  • Guilherme Araujo Mattos
  • A/D converters with adaptive sampling using Wavelet Transform in the analog domain and D/A converter with polynomialreconstruction using Lipschitz coefficients.

  • Líder : SANDRO AUGUSTO PAVLIK HADDAD
  • MIEMBROS DE LA BANCA :
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • FERNANDO CHAVEZ PORRAS
  • JOSE EDIL GUIMARAES DE MEDEIROS
  • SANDRO AUGUSTO PAVLIK HADDAD
  • Data: 04-ago-2022


  • Resumen Espectáculo
  • This work presents a proposal to reduce data and energy consumed in the signal conversion process, in the digital and analog domains. For this, twoconverter systems are used. An analog-digital (A/D) system with information compression and adaptive sampling using the Lipschitz coefficient andthe properties of the Wavelet Transform. The Lipschitz coefficient characterizes the waveform at a given point. Using the exponent compresses theinformation, reducing the number of samples necessary for the conversion process, yet, maintaining a high resolution. Another system developed inthis work is the dedicated digital-to-analog (D/A) converter, which performs the process of reconstruction of the analog signal by a polynomialarrangement. The signal is reconstructed with amplitude, time and exponent information. These are the A/D outputs, and consequently, the D/Ainputs. A reconstruction based on the morphology of the signal using a polynomial approximation method is used, significantly reducing the samplingrate.Circuits were developed that perform the calculation of the exponent using the Transform Wavelet. After building the necessary circuits in theVirtuoso tool, four waveforms were tested with known exponents, in order to verify that the system can perform detection. There was then a variationcircuit, temperature and power parameters, generating 300 samples for each signal. It was found statistically through the T test for the signs thatshowed a normal distribution, and using the test of Wilcoxon for the signal that did not present it. The results obtained were exponents that did notdeviate considerably from the value nominal, indicating that it is possible to obtain reliability in obtaining the exponent.In the construction of the D/A system, it was necessary to generate four base waves by combining and weighting them to generate any exponent n.In this work, 16 exponents were tested (4 bits in the exponent information) spaced in different ways, varying the value between 2 and 0.25. Thewaves obtained by the designed circuits were tested comparing them with the ideal waveforms. The system output shows an RMS error of less than636 uV for the steepest curve. The reconstruction of an ECG signal was also performed, where there is a reduction in the sampling rate of 95.08% inrelation to an ideal converter with a linear sampling rate of 2 KS/s. A high reduction in the information needed for the reconstruction, even whenperforming a comparison with a converter that has a low sampling rate.

2
  • Jeronimo da Silva Avelar Filho
  • Detecção de Esquizofrenia com Base em Análise Estrutural do Cérebro Usando Aprendizagem de Máquina Aplicada a Combinações de Cortes em Imagens Volumétricas de Ressonância Magnética

  • Líder : CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MIEMBROS DE LA BANCA :
  • ADSON FERREIRA DA ROCHA
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • FABRICIO ATAIDES BRAZ
  • JOALBO MATOS DE ANDRADE
  • Data: 09-ago-2022


  • Resumen Espectáculo
  • Schizophrenia is a mental disease with many clinical manifestations, making the diagnosis a significant challenge. Until a correct diagnosis is attained, the patient experiments with mental suffering that can lead to social conflicts, involuntary accidents, and suicides. Despite the clinical complexity, early diagnosis is of utmost importance, and several recent studies focus on analyzing structural brain modifications that have been correlated to schizophrenia and can be detected in anatomical magnetic resonance images. Previous research applying machine learning to such images presented promising results. However, the scope was limited to analyzing only one or few slices of the brain while not using recent algorithms at the core of the classifiers. Furthermore, using fewer slices and simple algorithms can lead to information loss due to sub-optimal feature extraction. In this study, we created machine learning models based on Convolutional Neural Networks (CNN) and evaluated the best training parameters based on a Magnetic Resonance Images (MRI) dataset from schizophrenia-diagnosed patients and a subjects control group. In addition, we analyzed the performance of the classifiers, first trained with individual slices of the brain and later with different combinations of multiple magnetic resonance slices. We obtained the slices by extracting them one by one from the scanned volume output of the MRI scanning process. First, the slices were numbered based on the axial index number of the scanned volume. Then, we experimented with selecting the slices using metrics like covariance and entropy. The best results were obtained when we used the entropy concept to evaluate the slices images. The slices were sorted by the greatest entropy, and using this criterion, we evaluated each slice individually, using a machine learning model and the collections of slices. Our strategy was to create datasets with incremented number of slices ordered by the entropy. First, we started training with a dataset containing only one individual slice from the scanned volumes. Then, in a second step, we used two slices to build the dataset, and so on, until we created a dataset with all the images extracted from the volume. Each dataset, created from combinations of slices, was used to train the ML model and was evaluated to obtain the performance indicators. Our results suggest that it is possible to obtain an accuracy near 80% when trained with a previously selected combination of slices. Also, we explore the use of Explainable Artificial Intelligence (XAI) to comprehend the model output classification.

3
  • Thiago Alves Espindola
  • Learning Machine for automating classification of photographic records of Foot Diabetic Ulcers according to the Classification of the University of Texas.

  • Líder : MARCELLA LEMOS BRETTAS CARNEIRO
  • MIEMBROS DE LA BANCA :
  • JOSÉ CARLOS TATMATSU ROCHA
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • Data: 02-sep-2022


  • Resumen Espectáculo
  • Diabetic Foot Ulcer (DUP) is a complication of Type 1 and 2 Diabetes Mellitus (DM) characterized by wounds associated with ischemia, neuropathy, and deformities that can lead to a clinical picture of low probability of healing. The conventional method of treatment is based on dressings and drug intervention. Notably, assertive treatment depends on an accurate diagnosis, capable of directing care according to the individual need of the patient. The diagnosis, in turn, results from an assessment that is essentially based on the health professional's experience. In this sense, the present work proposes to present a diagnostic aid tool. Objective: To develop an algorithm that classifies UPD through photographic records, according to the parameters of the University of Texas Classification, using the UPD records obtained in the RAPHA® protocol as a sample, using latex biomembranes, with and without curcumin. Methodology: The writing of the algorithm used the Python language because it supports AI. Thus, the Machine Learning (AM) technique was selected, since it allows AI to work from learning based on examples (samples) in order to obtain a desirable result. Results: In the end, an initial version of the classifier was obtained, with preliminary tests of execution and validation of results carried out by health professionals. The results obtained proved to be satisfactory, according to the analysis of accuracy generated by the program itself and confirmed by health professionals, it had assertiveness between 98% and 100%, however, it needs refinement to optimize performance.

4
  • JOÃO CARLOS WOHLGEMUTH
  • STUDY OF MATERIALS TO ATTENUTE PULSING MAGNETIC FIELD GENERATOR OF INTERFERENCE IN PACEMAKERS.

  • Líder : VERA REGINA FERNANDES DA SILVA MARAES
  • MIEMBROS DE LA BANCA :
  • MILTON ERNESTO ROMERO ROMERO
  • RONNI GERALDO GOMES DE AMORIM
  • RUDI HENRI VAN ELS
  • VERA REGINA FERNANDES DA SILVA MARAES
  • Data: 15-sep-2022


  • Resumen Espectáculo
  • Pacemakers are formidable equipment, and an expressive number of people depend on them, with different levels of vital commitment, but they are exposed to electromagnetic interference from the most varied sources. Among them, an increasingly present type is the one produced by anti-theft systems in stores, making it a type of interference difficult to avoid, since they use antennas positioned at the access doors of commercial establishments. As they make use of the emission and reception of electromagnetic waves, these systems can affect the functioning of the pacemakers, causing discomfort or even putting the lives of users at risk. Studies have shown that the Magnetic Acoustic type is more likely to alter the functioning of pacemakers. Pacemaker manufacturers reserve the right to only communicate that their equipment meets the international standard ISO 14117 (Active Implantable Medical Devices – Electromagnetic Compatibility – Electromagnetic Compatibility Test Protocols for Implantable Cardiac Pacemakers, Implantable Defibrillators and Cardiac Resynchronization Devices), Updated in 2019, which defines in item 4.8 Protection Against Exposure to AC Magnetic Fields in the Range from 1 Hz to 140 kHz, an intensity of 150 A/m to which implantable cardiac devices must be exposed without presenting changes in their functioning. Manufacturers of goods security systems do not report the power of their equipment, but the SPCICED TEAS study carried out in 1998, with the participation of the main American manufacturers of these systems, reported that they measured a magnetic field strength of 310 A/m close to the broadcasting antennas. Therefore, this study explored the possibility of attenuating magnetic fields generated at the same frequency used by these magnetic acoustic anti-theft systems through silicon steel and aluminum plates, which have shown to be promising in the manufacture of personal shields. Silicon steel showed considerable shielding power even at this frequency, but was surpassed by aluminum in all positions where at least one face was transverse to the magnetic flux lines. The results provided great support for the development of shields to attenuate high frequency magnetic fields.

5
  • Luciene Ferreira dos Anjos
  • THE SELF-CARE PROCESS IN THE TREATMENT PROTOCOL OF FOOT DISEASES IN DIABETES: THE CASE OF THE RAPHA ® EQUIPMENT.

  • Líder : MARIO FABRICIO FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • MARIO FABRICIO FLEURY ROSA
  • ADSON FERREIRA DA ROCHA
  • ALDIRA GUIMARAES DUARTE DOMINGUEZ
  • JOSÉ CARLOS TATMATSU ROCHA
  • Data: 29-sep-2022


  • Resumen Espectáculo
  • INTRODUCTION: DIABETES MELLITUS (DM) IS DEFINED AS A SET OF MULTI-ETHNIC METABOLIC DISORDERS CAUSED BY THE LACK OF INSULIN AND/OR THE INABILITY OF INSULIN TO EXERCISE FULL EFFECT, WHOSE MAIN CHARACTERISTICS IS HYPERGLYCEMIA, THAT IS, THE “EXCESS OF SUGAR IN THE BLOOD”. THIS CONDITION MAY RESULT IN MACRO AND MICROVASCULAR COMPLICATIONS, RESULTING IN A FRAMEWORK OF KIDNEY INSUFFICIENCY, RETINOPATHY (VISION COMPROMISE), INFARCTION, STROKE, LOSS OF SENSITIVITY IN THE LOWER LIMBS, among others. IT IS CURRENTLY CONSIDERED A PUBLIC HEALTH PROBLEM, DUE TO THE GROWING NUMBER OF AFFECTED AND HIGH PUBLIC INVESTMENTS IN THE TREATMENT OF THE DISEASE. IN THIS SCENARIO, COMPLICATIONS RELATED TO THE LOWER LIMBS, CALLED “DISEASES OF THE FOOT IN DIABETES” ARE HIGHLIGHTED BY THE DIFFICULTY OF HEALING, WHICH CAN EXTEND THE TREATMENT FOR YEARS, AND ARE THE BIGGEST CAUSING FACTOR OF AMPUTATION OF THE LIMBS. AT THIS juncture, RESEARCHERS AT THE UNIVERSITY OF BRASÍLIA (UNB) DEVELOPED A PORTABLE TECHNOLOGY THAT ASSOCIATES THE USE OF NATURAL LATEX BIOMEMBRANE (OF THE RUBBER TREES HEVEA BRASILIENSE) AND LED PHOTOTHERAPY TO TREAT WOUNDS IN THE LOWER LIMBS OF DIABETIC PATIENTS, CALLED RAPHA® MEANS TO HEAL IN HEBREW. HAVING PROMISING RESULTS IN CLINICAL STUDIES, THE TECHNOLOGY IS BASED ON AN IN-HOUSE TREATMENT, PERFORMED BY THE PATIENT AND THEIR FAMILY, DISTANCE THE PATIENT FROM HEALTH CENTERS AND STIMULATE SELF-CARE. OBJECTIVE: TO ANALYZE THE PROCESS OF SELF-CARE AS A THERAPEUTIC STRENGTH IN THE SET OF THE RAPHA® PROTOCOL. METHODOLOGY: THIS RESEARCH IS SUPPORTED IN THE THEORETICAL METHODOLOGICAL FRAMEWORK OF QUALITATIVE RESEARCH. SUCH METHOD IS A TOOL THAT ALLOWS ACCESS TO THE STUDY PHENOMENON IN A SUBJECTIVE CONTEXT. THUS, WITH THE ASSUMPTION THAT NURSING OPERATES DURINGLY IN THE SELF-CARE PROCESS, A QUALITATIVE LOOK LINKED TO EVENTS IS JUSTIFIED. THE QUALITATIVE ANALYSIS WAS BASED ON PARTICIPANT OBSERVATION AND DOCUMENTAL ANALYSIS OF DATA COLLECTED DURING THE CLINICAL PHASE OF THE RAPHA® PROJECT, IN THE YEARS 2018 AND 2019. RESULTS: THE INTEGRATION OF THE SELF-CARE PROCESS INTO THE CLINICAL TEST (Once RAPHANO WHILE TECHNIQUES DEPENDS ON THE ASSERTIVE CONDUCT OF THE PATIENT) PROVIDED PARTICIPANTS A GREATER UNDERSTANDING OF THE HEALTH/DISEASE PROCESS AND DYNAMICS OF CARE, SUCH AS WOUND CLEANING. IN THE SAME WAY, IT STIMULATES PARTICIPATION AND SUPPORT OF FAMILY IN THE PROCESS, WHICH IMPACTS THE PARTICIPANT'S QUALITY OF LIFE. IT IS IMPORTANT TO UNDERSTAND THAT AS IT IS A HOME TREATMENT, REMOVAL FROM HEALTH CENTERS OPERATES POSITIVELY ON THE RESULTS, SINCE IT REDUCES THE RISK OF CONTACT WITH OTHER DISEASES AND COSTS WITH TRAVELING TO THE HEALTH CENTER. FINALLY, SELF-CARE PROVIDE TOOLS IN ADDITION TO CLOSING THE WOUND TO AVOID THE EMERGENCE OF NEW TOOLS, THROUGH DAILY CARE, WHICH IMPLIES SAVINGS WITH TREATMENT AND MEDICATION. CONCLUSION: THE OUTCOMES ACHIEVED IN THE RAPHA® RESEARCH DEMONSTRATED IN THE STATISTICAL DATA INDICATE THAT SELF-CARE CAN BE CONSIDERED A THERAPEUTIC STRENGTH, HAVING ACHIEVED RESULTS BIGGER THAN SUS COVERAGE IN CARE.

6
  • MANOEL DE JESUS ALMEIDA JUNIOR
  • ANALYSIS OF THE ECONOMIC EFFICIENCY OF DIABETIC FOOT TREATMENT IN THE UNITED HEALTH SYSTEM: A SYSTEMATIC REVIEW.
  • Líder : MARIO FABRICIO FLEURY ROSA
  • MIEMBROS DE LA BANCA :
  • MARIO FABRICIO FLEURY ROSA
  • MARCELLA LEMOS BRETTAS CARNEIRO
  • SUELIA DE SIQUEIRA RODRIGUES FLEURY ROSA
  • JOSÉ CARLOS TATMATSU ROCHA
  • Data: 05-oct-2022


  • Resumen Espectáculo
  • Diabetes Mellitus (DM) is a metabolic, chronic, non-communicable disease, with heterogeneous etiopathogenesis, triggered by defects in insulin secretion and/or action, causing disorders in carbohydrate, fat and protein metabolism. It is considered a public health problem due to its prevalence, impact on morbidity and mortality, and its complications. The IDF registered 537 million individuals with DM in 2021, with 80% of this population living in low- and middle-income countries. Brazil is the fourth country in the world in the number of DM cases, and 50% of diabetics are unaware of their diagnosis. Early diagnosis and initiation of treatment prevents and delays acute or chronic complications, micro and macrovascular, renal, and neurological; decreasing mortality and morbidity; and reducing the cost for families and the health system. The time of DM diagnosis is associated with increased risk of developing diabetic foot disease, one of the most frequent complications in these individuals. The health care expenditure for DM patients is $727 billion dollars annually, measured by the IDF in 2021. The objective of this study is to analyze the economic efficiency of the technologies made available by SUS for the treatment of wounds and foot ulcers in patients with type I and II diabetes mellitus. This research was conducted through a systematic review in which different sources were analyzed and discussed. As a result, 10 eligible articles were screened, 5 of which were RCTs, for this study and in this analysis it was observed that telemedicine in the treatment of the diabetic foot was effective and generated significant savings to the health system; the human dehydrated amnion and chorion allograft showed better cost-utility when compared to the standard isolated treatment of the diabetic foot, generating savings. A significant economic and health burden for the health system was observed in the analyzed studies, which highlights the urgency of public policies aimed at prevention. It was also observed the need to identify the best health technologies and their cost-effectiveness for the public health system. Conclusion: Telemedicine proved to be efficient in rapidly screening diabetic foot ulcers, was effective in the course of treatment, and proved to be economically feasible and less costly when compared to outpatient care. With respect to treatment, the dehydrated human amnion and chorion allograft showed cost-effectiveness compared to treatment alone. Further economic studies and intervention projects are needed to insert these changes in the health system permanently improving the treatment and prognosis of patients with diabetic foot ulcer, and reducing the costs related to these individuals in the SUS.

7
  • JOSELMA DA SILVA GUIMARAES
  • Proposal and Implementation of a New Head Injury Criterion Using the Hybrid III 50% Dummy, Including the Effects of Angular Speeds and Accelarations in Frontal Vehicle Collisions

  • Líder : CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • MIEMBROS DE LA BANCA :
  • ALESSANDRO BORGES DE SOUSA OLIVEIRA
  • CRISTIANO JACQUES MIOSSO RODRIGUES MENDES
  • Francisco Vieira Garonce
  • ROBERTO DE SOUZA BAPTISTA
  • Data: 07-nov-2022


  • Resumen Espectáculo
  • Data from the National Registry of Accidents and Traffic Statistics (RENAEST) provide important alerts regarding automobile accident rates, and the resulting numbers of victims, injuries, and deaths in the last four years, in Brazil. Despite a 50\% reduction in the accident indicators per 100 thousand inhabitants, registered in the last 10 years, the numbers reported by the organizations are worrisome. Associated with the number of victims in traffic accidents, one can emphasize the percentage of victims with head lesions. According to the World Health Organization (WHO), head injuries are the leading cause of death and debilitating trauma in motor vehicle users.

    Today, international government agencies and the New Car Assessment Program (NCAP) are widespread and conduct the important work of submitting new cars to a wide range of crash tests, to provide consumers with independent information regarding the safety levels of marketed vehicle models. These programs use injury criteria for different parts of the body, such as the head injury criterion (HIC). Despite the HIC's importance in quantifying one of the aspects of legion risks for a given vehicle, we observed some limitations in its central concept, as well as a shortage of specific studies about correlated aspects. For example, HIC includes a translational acceleration component in its computation but does not include angular acceleration components. Also, there is a HIC threshold below which the vehicle is considered safe against severe injuries (threshold of 700, according to NCAP), but we did not find studies about light and moderate lesions for HIC values below such threshold.

    In this context, we propose the definition, implementation, and evaluation of a new head injury criterion that takes into account not only the translational acceleration but also rotational moment components, i.e., the centripetal and tangential components. The proposed aspect is that the new index, which we call the modified HIC, should evaluate aspects that are already present in the standard HIC, but also rotational HIC components that are not predicted by that standard.

    To evaluate the proposed index, we considered acceleration and force signals obtained in dummies during frontal collisions. We compared the results from the new index to those existing criteria, and the modified HIC proved more sensitive than the standard HIC in predicting brain lesion risk. The study also predicts the possibility of use in the Male Dummy Hybrid III 50\%, instead of the MALE THOR 50\% model when computing angular velocities and accelerations. This can reduce the costs of collision tests for evaluating the proposed index.

8
  • Rafael Henrique Santos de Brito
  • INNOVATIVE TREATMENTS FOR CYSTIC FIBROSIS: A SYSTEMATIC REVIEW

  • Líder : RONNI GERALDO GOMES DE AMORIM
  • MIEMBROS DE LA BANCA :
  • MARCOS DE VASCONCELOS CARNEIRO
  • MARILIA MIRANDA FORTE GOMES
  • RONNI GERALDO GOMES DE AMORIM
  • VERA REGINA FERNANDES DA SILVA MARAES
  • Data: 29-nov-2022


  • Resumen Espectáculo
  • Introduction: Cystic Fibrosis is an autosomal-recessive and multisystemic disease, resulting from mutations in the seventh human chromosome at position 7q13, pathogenic variants in both alleles of the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene, which encodes a protein of the same name, responsible by encoding the transmembrane conductance regulatory protein of chlorine, causing dysregulation of electrolyte levels in the body's cells, often considered a severe disease that greatly impacts the respiratory system.
    Objective: To study the current picture of the treatment of this disease and the conditions for improving the daily quality of life of patients treated with CFTR dysfunction modulators.
    Methodology: Initially, an exploratory literature review was performed followed by a systematic review guided by the PICO strategy, acronym for Patient, Intervention, Comparison and “Outcomes”, translated as Outcome.
    Conclusion: CFTR modulating drugs (ivacaftor, lumacaftor and tezacaftor), administered as monotherapy or in combination, are revolutionizing the treatment of CF. Although they are not curative, these therapies already represent prospects for prolonging and increasing the quality of life of the CF population, mainly reducing the number of pulmonary exacerbations.

SIGAA | Secretaria de Tecnologia da Informação - STI - (61) 3107-0102 | Copyright © 2006-2024 - UFRN - app33.sigaa33