| Sumario: | Background: Global demographic trends toward aging populations highlight growing mental health challenges among older adults. Traditional methods for assessing psychological well-being are often infrequent, inaccessible, and subject to self-report biases, underscoring the need for innovative monitoring solutions. Research aim: The paper presents a proof-of-concept design for a multimodal chatbot- based system intended to support continuous monitoring of psychological wellbeing in older adults. It explores how natural language processing (NLP) and speech analysis technologies can be integrated to identify emotional states and enable proactive support. Methods: The proposed system integrates daily conversational interactions via an empathetic chatbot, employing automated textual sentiment analysis through a finetuned Large Language Model (LLM) and prosodic speech analysis to assess emotional states. These analyses are complemented by regular subjective self-assessments to create comprehensive and personalized well-being profiles. Results: The developed prototype shows the conceptual feasibility of unobtrusive monitoring and early detection of mental distress. The proposed design further includes mechanisms for personalized recommendations and triggering interventions when significant declines in well-being are detected. Conclusions: The current proof-of-concept requires future empirical validation of its usability, reliability, and accuracy. Further research should also address technical refinements, ethical considerations, and broader cultural adaptability.
|
|---|