Offline Policy Evaluation of Multi-Turn LLM Health Coaching with Real Users

Melik Ozolcer, Sang Won Bae

Keywords: LLM agents, health coaching, offline policy evaluation

TL;DR: Deployed LLM health coach: OPE shows heavy-tool policies boost averages but harm some subgroups; a brief early information-gain phase improves personalization, pass@3, and goal success

Abstract: We study a web-deployed, tool-augmented LLM health coach with real users. In a pilot with seven users (280 rated turns), offline policy evaluation (OPE) over factorized decision heads (Tool/Style) shows that a uniform heavy-tool policy raises average value on logs but harms specific subgroups, most notably low-health-literacy/high-self-efficacy users. A lightweight simulator with hidden archetypes further shows that adding a small early information-gain bonus reliably shortens trait identification and improves goal success and pass@3. Together, these early findings indicate an evaluation-first path to personalization: freeze the generator, learn subgroup-aware decision heads on typed rewards (objective tool outcomes and satisfaction), and always report per-archetype metrics to surface subgroup harms that averages obscure.

Submission Number: 140