2026-03-132026-03-132026-02-27WANZELER, Gabriele Rodrigues. Desenvolvimento de um protótipo para triagem automatizada de glaucoma por deep learning integrado ao WhatsApp. Orientador: Otávio Noura Teixeira. 2026. [27] f. Trabalho de Curso (Bacharelado em Engenharia da Computação) – Faculdade de Engenharia da Computação, Campus Universitário de Tucuruí, Universidade Federal do Pará, Tucuruí, 2026. Disponível em: https://bdm.ufpa.br/handle/prefix/9343. Acesso em:.https://bdm.ufpa.br/handle/prefix/9343This work evaluates a ResNet50 model with transfer learning for automated glaucoma detection in fundoscopy images, integrated into a WhatsApp prototype for healthcare professionals. The model was trained on the ACRIMA and RIM-ONE datasets. On ACRIMA, it achieved 98.28% sensitivity, F1-score of 0.982, and AUC of 0.992; on RIM-ONE, 83.33% sensitivity, F1-score of 0.869, and AUC of 0.949, reflecting class imbalance. The prototype achieved accuracy rates of 83.33% and 80%, with average confidence of 94.80% and 87.31%, respectively. Results show that data balance directly influences performance, reinforcing the potential of deep learning as a support tool for glaucoma screening.Acesso AbertoAprendizado profundoRedes neurais convolucionaisVisão computacionalDeep learningGlaucomaCNPQ::ENGENHARIASCNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::METODOLOGIA E TECNICAS DA COMPUTACAOTrabalho de Curso - Graduação - ArtigoAttribution-NonCommercial-NoDerivs 3.0 Brazil