SePA: A Search-enhanced Predictive Agent for Personalized Health Coaching

2025-09-04

AI Summary

Existing AI health coaches suffer from insufficient personalization and lack of evidence-based reliability. Method: We propose the Search-Augmented Predictive Agent (SAPA), a framework integrating personalized time-series forecasting models with retrieval-augmented generation (RAG) to dynamically fuse streaming wearable sensor data and expert-validated, authoritative medical knowledge bases. SAPA employs rolling-origin and grouped cross-validation to ensure generalizability. Results: Personalized models significantly outperform generic baselines; expert blind evaluation confirms clinically meaningful health recommendations (Cliff’s δ = 0.3, p = 0.05); and SAPA achieves a quantifiable trade-off between response quality and latency. This work establishes a reproducible methodology and empirical foundation for trustworthy, adaptive, and evidence-based AI health coaching systems.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This paper introduces SePA (Search-enhanced Predictive AI Agent), a novel LLM health coaching system that integrates personalized machine learning and retrieval-augmented generation to deliver adaptive, evidence-based guidance. SePA combines: (1) Individualized models predicting daily stress, soreness, and injury risk from wearable sensor data (28 users, 1260 data points); and (2) A retrieval module that grounds LLM-generated feedback in expert-vetted web content to ensure contextual relevance and reliability. Our predictive models, evaluated with rolling-origin cross-validation and group k-fold cross-validation show that personalized models outperform generalized baselines. In a pilot expert study (n=4), SePA's retrieval-based advice was preferred over a non-retrieval baseline, yielding meaningful practical effect (Cliff's $δ$=0.3, p=0.05). We also quantify latency performance trade-offs between response quality and speed, offering a transparent blueprint for next-generation, trustworthy personal health informatics systems.

Problem

• Integrates personalized machine learning for health coaching
• Combines predictive models with retrieval-augmented generation
• Ensures evidence-based guidance using expert-vetted content

Innovation

• Personalized machine learning from wearable data
• Retrieval-augmented generation with expert-vetted content
• Cross-validated predictive models outperforming generalized baselines