HEART (Health Enhancement and Analysis Resource Tool): An AI-Based Platform Integrating Validated Assessments and Personalized Recommendations for Healthy Lifestyles

Objectives: Digital tools are increasingly used to improve lifestyle. In this context, a web-based platform was developed utilizing artificial intelligence (AI) and retrieval-augmented generation (RAG) to deliver personalized lifestyle recommendations exclusively based on a knowledge base manually selected and validated by academic experts. Methods: The platform comprises an expert-led content management backend and an AI-driven user interface. Peer-reviewed articles and clinical guidelines were selected and uploaded by organizing the content into three domains Physical, Mental, and Social. Validated questionnaires were integrated to both improve user profiling and scoring lifestyle. User behaviour and questionnaire scores are analyzed by the AI to provide personalized guidance while ensuring all responses are strictly derived from the validated knowledge base. Results: A functional prototype was developed. Internal technical validation confirmed that the RAG system restricted knowledge retrieval to the expert-uploaded documents, with no information retrieval from external, unverified sources. Initial expert review confirmed that AI-generated recommendations were strictly consistent with the uploaded evidence and validated tools. The integration of validated tools with the RAG engine enabled the system to generate proactive, personalized lifestyle suggestions while anchoring responses to the repository. Conclusions: HEART enables the evaluation of user lifestyle and development of personalized guidance by translating validated tools and evidence-based guidelines into digital systems. The platform provides a structured and controlled environment for the assessment of lifestyle and promotion of health by grounding AI-driven outputs in validated instruments. Future research will be planned to evaluate outcomes and usability. Funding Sources: European Union-NextGenerationEU under the Italian Ministry of University and Research (MUR) National Innovation Ecosystem AI Disclosures: Evidence-based responses are ensured by a Retrieval-Augmented Generation architecture. For document retrieval, the employed vector database is Pinecone. For natural language understanding and text generation, OpenAI language models are used. Authors have reviewed the content and assume full responsibility for the accuracy and integrity of the content presented.