| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 9.29 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
A ansiedade é uma das condições de saúde mental mais prevalentes da atualidade, afetando
milhões de pessoas e comprometendo o seu bem-estar e funcionalidade no quotidiano. Este
contexto tem motivado o crescimento de intervenções digitais baseadas em tecnologia, que se
têm demonstrado eficazes na promoção da literacia emocional, prevenção de sintomas e
estímulo à autorregulação emocional. Estas intervenções, embora não substituam
acompanhamento clínico, têm revelado utilidade como suporte complementar, especialmente
junto de públicos com acesso limitado a cuidados especializados.
O objetivo principal desta dissertação foi desenvolver e avaliar um assistente pessoal contextual
inteligente, capaz de apoiar a prevenção e redução da ansiedade, fornecendo recomendações
e intervenções não clínicas, adaptadas ao estado emocional e ao contexto do utilizador. A
aplicação propõe-se assim como uma ferramenta de autogestão emocional personalizada,
acessível e contínua.
A metodologia seguiu uma abordagem iterativa centrada no utilizador, iniciando-se com uma
revisão de aplicações e literatura científica. Com base nessa análise, foi concebido um sistema
digital com funcionalidades como registo de humor, gestão de tarefas e agenda, notificações
inteligentes e um chatbot contextual, suportado por um modelo de linguagem natural (LLM) e
uma base vetorial (ChromaDB) para recuperação semântica de conteúdos.
A avaliação envolveu 10 participantes, com recolha de dados quantitativos e qualitativos. Para
além da aceitação e usabilidade, foram analisados níveis de ansiedade através da escala GAD-
7, e consciência emocional percebida por via de questionários e feedback direto. Os resultados
demonstram uma aceitação elevada da aplicação, com destaque para a sua utilidade na
organização pessoal, registo emocional e incentivo a práticas de autocuidado. Os dados
sugerem ainda que a ferramenta contribuiu para um maior reconhecimento dos estados
emocionais e redução de sintomas de ansiedade, especialmente em momentos de maior
vulnerabilidade.
Conclui-se que a integração de inteligência artificial com personalização contextual representa
um contributo relevante e promissor para a saúde mental digital, oferecendo uma solução
empática, eficaz e tecnicamente robusta, com potencial de evolução e aplicação em contextos
mais amplos.
Anxiety is one of the most prevalent mental health conditions worldwide, significantly affecting individuals' well-being and daily functioning. In response, there has been a growing interest in digital technology-based interventions, which have shown effectiveness in emotional literacy, symptom prevention, and self-regulation support. While these solutions are not clinical treatments, they have proven useful as complementary tools, especially for populations with limited access to specialized mental health care. The main objective of this dissertation was to develop and evaluate a context-aware digital assistant designed to support anxiety prevention and reduction by providing non-clinical, personalized interventions and recommendations, tailored to each user’s emotional state and context. The application aims to function as a continuous, accessible tool for emotional selfmanagement. The methodology followed an iterative, user-centred approach, beginning with a systematic review of relevant applications and scientific literature. Based on this analysis, a digital system was designed, featuring mood tracking, task and agenda management, intelligent notifications, and a contextual chatbot, powered by a large language model (LLM) and a vector database (ChromaDB) for semantic information retrieval. The evaluation involved 10 participants, with data collected through both quantitative and qualitative methods. In addition to analysing usability and user acceptance, the study measured anxiety levels using the GAD-7 scale, and perceived emotional awareness through questionnaires and direct feedback. Results showed high user acceptance, with participants highlighting the system’s usefulness for daily organization, emotional tracking, and promoting self-care habits. The data also suggest an improvement in emotional awareness and a reduction in anxiety symptoms, particularly during periods of emotional vulnerability. In conclusion, integrating artificial intelligence with contextual personalization offers a promising contribution to digital mental health, providing an empathetic, effective, and technically robust solution with the potential for broader application and long-term development.
Anxiety is one of the most prevalent mental health conditions worldwide, significantly affecting individuals' well-being and daily functioning. In response, there has been a growing interest in digital technology-based interventions, which have shown effectiveness in emotional literacy, symptom prevention, and self-regulation support. While these solutions are not clinical treatments, they have proven useful as complementary tools, especially for populations with limited access to specialized mental health care. The main objective of this dissertation was to develop and evaluate a context-aware digital assistant designed to support anxiety prevention and reduction by providing non-clinical, personalized interventions and recommendations, tailored to each user’s emotional state and context. The application aims to function as a continuous, accessible tool for emotional selfmanagement. The methodology followed an iterative, user-centred approach, beginning with a systematic review of relevant applications and scientific literature. Based on this analysis, a digital system was designed, featuring mood tracking, task and agenda management, intelligent notifications, and a contextual chatbot, powered by a large language model (LLM) and a vector database (ChromaDB) for semantic information retrieval. The evaluation involved 10 participants, with data collected through both quantitative and qualitative methods. In addition to analysing usability and user acceptance, the study measured anxiety levels using the GAD-7 scale, and perceived emotional awareness through questionnaires and direct feedback. Results showed high user acceptance, with participants highlighting the system’s usefulness for daily organization, emotional tracking, and promoting self-care habits. The data also suggest an improvement in emotional awareness and a reduction in anxiety symptoms, particularly during periods of emotional vulnerability. In conclusion, integrating artificial intelligence with contextual personalization offers a promising contribution to digital mental health, providing an empathetic, effective, and technically robust solution with the potential for broader application and long-term development.
Descrição
Palavras-chave
Anxiety Mental Health Artificial Intelligence Recommender Systems Personalization Digital Applications Ansiedade Saúde mental inteligência artificial sistemas de recomendação Personalização Aplicações digitais
