Banca de DEFESA: Joana de Albuquerque Ribeiro
Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.STUDENT : Joana de Albuquerque Ribeiro
DATE: 28/02/2023
TIME: 14:00
LOCAL: NÚCLEO DE MEDICINA TROPICAL
TITLE:
ECTODEX, an electronic identification key to ectoparasites of public-health interest: development and pilot evaluation
KEY WORDS:
Application, Android/iOS, wingless vectors, morphology, taxonomy, environmental surveillance, ectoparasitoses
PAGES: 124
BIG AREA: Ciências da Saúde
AREA: Medicina
SUBÁREA: Clínica Médica
SPECIALTY: Doenças Infecciosas e Parasitárias
SUMMARY:
Although critical for the efficient control-surveillance of several major infectious diseases (including plague, spotted fever, or Lyme disease), identification of ectoparasitic arthropods often relies on outdated, incomplete, and hard-to-access printed keys. Aiming to facilitate ectoparasite identification, we developed an Android/iOS app-based pictorial, polytomous key (“EctoDex”) to 34 species of ticks, lice, fleas, and bedbugs of public-health interest. In a pilot evaluation, we compared EctoDex with a printed dichotomous key (“PKey”) in terms of (i) percentage of correct identifications (“accuracy”) and (ii) time taken to complete an identification task (“time”). Moreover, we tested whether and how performance was affected by user traits (e.g., age, gender, background training, or prior expertise) and ectoparasite species. Each of 33 Brazilian users received up to 30 coded ‘problem specimens’ (21 species) preserved in tubes or on microscope-slides, and was asked to identify each specimen to species using EctoDex and PKey (1356 identification tasks in total). After exploratory analyses, we fitted generalized linear mixed models (GLMMs) accounting for dependencies among repeated observations by the same user; of the same specimen; and from same-genus ectoparasites. Our analyses show that EctoDex overall improved ectoparasite identification across users and species, both in terms of accuracy and time, relative to PKey. Accuracy improved by a larger amount among users without specialized training in ectoparasite taxonomy, with GLMM-predicted values rising from ~57.2% (PKey; CI95% [40.8–72.2]) to ~77.1% (EctoDex; [63.5–86.8]), than among users with such specialized training – whose accuracy rose only slightly, from ~73.1% (PKey; [60.4–82.9]) to ~74.4% (EctoDex; [62.0–83.9]). Identifications by users with high prior expertise were overall more accurate (EctoDex: ~87%; PKey: ~81%) than those by non-expert users (~59% and ~47%, respectively); in contrast, user age, gender, or background general training did not significantly affect accuracy with either key. Accuracy varied widely across ectoparasite species, ranging from 12–18% for the tick, Amblyomma parvum to ~98–99% for the louse, Pediculus humanus. Time GLMMs also showed that EctoDex speeded-up identification tasks particularly for users without specialized training in ectoparasite taxonomy (PKey: ~2.8 min/task, [2.3–4.0]; EctoDex: ~2.0 min/task, [1.4–2.8]); time-saving was modest for users with such training (PKey: ~2.5 min/task, [1.9–3.3]; EctoDex: ~2.1 min/task, [1.6–2.8]). Again, high prior expertise also led to large time-savings (~1.5 min/task, on average, vs. ~3.7 min/task for non-experts), and user age, gender, or background general training did not significantly affect time-to-identification. We also found large among-species variation, with average predicted values ranging from ~3–4 min/task for A. sculptum to ~1 min/task for P. humanus. We conclude that EctoDex holds promise as a means to improve ectoparasite identification in terms of both accuracy and time-savings. Improvements were larger for users lacking formal specialized taxonomic training – whose overall performance, when using EctoDex, became comparable to that of formally trained users. Our pilot study also highlighted, however, the substantial difficulties inherent to identifying some ectoparasites of public-health interest, including several tick species in the genus Amblyomma.
BANKING MEMBERS:
Presidente - 3343228 - RODRIGO GURGEL GONCALVES
Interna - 3088161 - JULIANA LOTT DE CARVALHO
Externa ao Programa - 2476936 - MARINA REGINA FRIZZAS
Externo à Instituição - CLAUDIO MANUEL RODRIGUES