Reproduce “SePA: A Search-enhanced Predictive Agent for Personalized Health Coaching” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Melik Ozolcer, Sang Won Bae
SePA integrates personalized predictive models for daily stress, soreness, and injury risk using wearable sensor data with a context-aware, trusted web-retrieval pipeline. This system generates proactive, individualized health coaching advice grounded in expert-vetted sources and provides verifiable citations, transforming reactive data reporting into preventive guidance.
Key insights
- Personalized health models (PHMs) are essential for accurately predicting subjective states like stress and injury risk, achieving R² values over 0.50 and 0.40 respectively, significantly outperforming generalized models which often show negative R².
- Integrating ML-driven personalized risk predictions into a context-aware Retrieval-Augmented Generation (RAG) pipeline substantially enhances the quality, relevance, and helpfulness of LLM-generated health advice.
- Grounding LLM responses in expert-vetted, whitelisted web sources with explicit citations is critical for establishing trust and transparency in AI-powered health coaching.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.