Multi agent ai improves personalized health checkup advice quality
A Multi-Agent LLM Framework for Personalized Health Checkup Interpretation and Guidance
Artificial IntelligenceMultiagent Systems
Summary
Interpreting health checkup results for one person can be complicated because it involves looking at medical records over time, understanding medical knowledge, and giving lifestyle advice. The authors created a system where multiple AI agents work together to answer complex health questions by dividing them into smaller tasks. They found that this multi-agent approach gave better answers that were more useful and consistent, especially when looking up personal health records. However, it took a bit more time and computing power to run compared to using a single agent.
What this means in practice
- •For digital health product teams: Integrate multi-agent AI to enhance personalized interpretation and guidance from health checkup data for users with complex queries.$Commercial implications: It enables development of advanced health recommendation tools that provide tailored feedback, improving user engagement and trust.
- •For medical chatbot developers: Use multi-agent frameworks to improve handling of compound patient questions by decomposing intents and providing consistent multi-faceted responses.
Authors
HyungJun Kim, Taehan Lee, Soojin Cheon
Abstract
Personalized interpretation of health checkup results requires reasoning across longitudinal records, medical knowledge, lifestyle guidance, and healthcare navigation. We present a multi-agent large language model (LLM) system that identifies multiple intents, maps each to a task-specific agent, executes them in parallel, and synthesizes their outputs. We compared answers generated in Single Agent and Multi Agent settings on 120 Korean compound queries combining two to four requirements, using synthetic health checkup records. The Multi Agent improved the weighted LLM-judge score from 1.695 to 1.797 (p = 0.027), and three additional LLM judges showed consistent improvements ($Δ$ = +0.111 to +0.186, all p < 0.05). The gains came from usefulness, consistency, and the handling of every requirement in compound queries, whereas numerical accuracy and grounding improved significantly under only one of the four judges and medical safety did not differ, and critical failures occurred at similar rates (Single Agent 15.0% vs. Multi Agent 13.3%). Two human evaluators preferred Multi Agent in 66.7% and 68.3% of pairwise comparisons. Multi Agent execution increased latency and cost by 1.31$\times$ and 2.02$\times$, respectively. In exploratory subgroup analyses, the improvement was concentrated in queries involving personal-record lookup.