An Environmental-AI Integrated Recommender System for Parkinson’s Rehabilitation: A Prototype Study in Taiwan

Abstract

This study presents a prototype Environmental-AI recommender system for Parkinson’s disease rehabilitation. Using Taiwan as a testbed, the system integrates real-time environmental data—including air quality, meteorological conditions, and satellite-derived indices —with individual health data such as symptom logs and wearable outputs. A rule-based engine interprets environmental conditions, while a generative AI model converts combined inputs into plain-language daily recommendations. The system supports both diagnosed Parkinson’s disease patients and high-risk individuals, offering symptom-aware or preventive lifestyle guidance. Simulated user profiles demonstrate how recommendations adapt to varying environmental and personal contexts. Results show the suggestions align with Parkinson’s disease care guidelines and environmental health principles. While promising, the system requires further validation, particularly to address AI hallucination risks and real-world clinical effectiveness. This approach illustrates the potential of combining environmental informatics and generative AI to support personalized chronic disease management.