Clincare: desenvolvimento de um assistente conversacional inteligente para apoio ao acompanhamento domiciliar de idosos em vulnerabilidade social por agentes comunitários de saúde

Abstract

This work presents the development of a software technical report aimed at strengthening Primary Health Care within the context of the Unified Health System. The project proposes an intelligent web platform to assist Community Health Workers and the monitoring of elderly patients in situations of social vulnerability. The solution integrates a voice-based conversational agent powered by large language models, specifically Llama 3.2 3B, using retrieval-augmented generation grounded in official Unified Health System protocols. Among the developed features, key highlights include intelligent triage with priority levels for home care visits, automatic intent classification, management of the elderly patient roster for the health worker, and an assisted registration workflow to mitigate digital exclusion among the elderly. The results demonstrate the technical feasibility of deploying artificial intelligence in resource-constrained environments, with average processing times of 6.75 seconds, providing a systematized mechanism for continuous risk identification and the improvement of communication between citizens and the health service.