A Self-Evolving Agent for Longitudinal Personal Health Management

2026-07-15

AI Summary

This work addresses the limitations of existing health AI systems, which typically operate through isolated interactions and struggle to adapt to users’ evolving health states and preferences over time. The authors propose HealthClaw, the first longitudinal personal health management agent architecture featuring a controlled self-evolving memory mechanism. Through a modular design that decouples shared safety rules, medical knowledge, and private longitudinal memory, HealthClaw dynamically updates user profiles and enables reusable interaction workflows while preserving accuracy, privacy, and safety. Empirical results demonstrate substantial improvements: across 900 longitudinal interactions, task accuracy increased from 0.2% to 45.7%, with a 71.7% reduction in prompt context exposure. Furthermore, HealthClaw achieved an average performance gain of 27.0 percentage points across nine biomedical tasks—seven of which remained statistically significant after multiple-testing correction—and exhibited superior privacy preservation in 100 dedicated privacy evaluations.

Abstract

Personal health management unfolds over repeated encounters, yet most health AI systems treat each request in isolation. We developed HealthClaw, an open-source agent architecture that updates support as a person's routines, preferences, measurements and risks change. It separates shared safety rules and medical knowledge from private longitudinal memory containing profile facts, reusable procedures and episodic traces. After each episode, induction determines what should update the profile, revise a procedure, remain episodic or be excluded. We evaluated HealthClaw with a synthetic year-long benchmark and nine 200-case biomedical tasks. Across 900 longitudinal support probes, answer accuracy increased from 0.2% with current-query prompting to 45.7% with HealthClaw, while prompt-side context exposure was 71.7% lower than with full-history prompting. In 100 privacy probes, HealthClaw produced higher privacy-aware answer quality and fewer unsafe disclosures than both baselines. Across the biomedical tasks, the mean absolute gain in the task-specific primary metric was 27.0 percentage points, and seven gains remained significant after false-discovery-rate correction. These offline benchmarks support governed, self-evolving memory for longitudinal personal health agents, although clinical effectiveness requires prospective evaluation. HealthClaw is publicly available at https://github.com/HC-Guo/HealthClaw.

Problem

• longitudinal health management
• personal health AI
• health agent
• continual learning
• privacy-aware health support

Innovation

• self-evolving agent
• longitudinal health management
• privacy-aware memory
• inductive memory update
• modular knowledge separation