Identificação de Padrões Multimodais de Comportamento Humano Utilizando Fenotipagem Digital

• Ano de defesa: 2023
• Autor(a) principal: MOURA, Ivan Rodrigues de
• Orientador(a): SILVA, Francisco José da Silva e
• Banca de defesa: SILVA, Francisco José da Silva e , COUTINHO, Luciano Reis , KON, Fábio , ENDLER, Markus , RABÊLO, Ricardo de Andrade Lira
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Maranhão
• Programa de Pós-Graduação: PROGRAMA DE PÓS-GRADUAÇÃO DOUTORADO EM CIÊNCIA DA COMPUTAÇÃO
• Departamento: DEPARTAMENTO DE INFORMÁTICA/CCET
• País: Brasil
• Palavras-chave em Português: Padrões comportamentais; Fenotipagem digital; Comportamento humano; Dispositivos ubíquos
• Palavras-chave em Inglês: Behavioral patterns; Digital phenotyping; Human behavior; Ubiquitous devices
• Área do conhecimento CNPq: Banco de Dados
• Link de acesso: https://tedebc.ufma.br/jspui/handle/tede/5260

Digital phenotyping is a research area that proposes the automatic collection of context data through sensors available in pervasive devices, allowing computational techniques to process this data to automatically detect human behaviors (e.g., sociability, physical activity). This information can support professionals specialized in monitoring and treating the health of individuals. Based on this scenario, this study proposes a solution capable of processing behavioral inference to recognize behavior patterns. These patterns are designed based on context attributes to model individuals’ behavior in specific situations, such as weekends and working days. Also, the proposed solution recognizes behavioral changes through knowledge modeling of the health specialist from fuzzy logic concepts. The preliminary experiments identified that the routine stability of individuals presents a high positive correlation with the solution’s ability to recognize multimodal behavioral patterns capable of modeling the behavioral routine. This evaluation also recognizes that the proposed solution is sensitive to identifying behavioral changes. Finally, we present an analysis of the influence of the solution’s hyperparameters on learning context-enriched human behavior patterns. Based on this analysis, we designed guidelines to support the parameterization process of the proposed solution.