Using Artificial Intelligence to Aid Depression Detection

• Autor(a) principal: Neves, Deângela Caroline Gomes
• Data de Publicação: 2019
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: eng
• Título da fonte: Repositório Institucional da UFRN
• Texto Completo: https://repositorio.ufrn.br/handle/123456789/43663
• topic: Deep Learning; Depression; DAIC-WOZ

Depression is a mental illness that affects a person’s mood, thinking, and behavior. Besides personal distress, depression is also considered a matter of public health. Recent research shows the advantages of using machine learning algorithms to automate and improve the screening for depression. In this thesis, we address the Distress Analysis Interview Corpus-Wizard of Oz (DAIC-WOZ) database, comprising clinical interviews and questionnaire assessments of over a hundred individuals. To automate the screening, we investigate a deep learning multimodal model, combining audio, visual, and text features. These features are extracted using VGGish model, OpenFace, and Doc2Vec, respectively, and fed into a multilayer perceptron (MLP) network to classify individuals as depressed or non-depressed. We compare the proposed approach to similar existing approaches from the literature through standard binary classification metrics.