Structured Exploration vs. Generative Flexibility: A Field Study Comparing Bandit and LLM Architectures for Personalised Health Behaviour Interventions

2026-03-06

AI Summary

Effectively selecting and delivering behavior change techniques (BCTs) to enhance digital health interventions remains challenging, particularly in achieving personalization and contextual adaptation. This study addresses this gap through a four-week field experiment (N=54) comparing five messaging strategies—random templates, a contextual multi-armed bandit, pure LLM-generated messages, a hybrid bandit+LLM approach, and an LLM incorporating interaction history—in promoting physical activity motivation. Results indicate that LLM-generated messages significantly outperform template-based approaches, though no significant differences emerge among LLM variants. While the bandit increases BCT diversity, it does not improve perceived usefulness. Qualitative analysis reveals that users value timely contextual responsiveness more than BCT optimization per se. The work proposes reflective AI design principles that balance structured exploration with generative flexibility.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Interaction Techniques and Devices
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Behaviour Change Techniques (BCTs) are central to digital health interventions, yet selecting and delivering effective techniques remains challenging. Contextual bandits enable statistically grounded optimisation of BCT selection, while Large Language Models (LLMs) offer flexible, context-sensitive message generation. We conducted a 4-week study on physical activity motivation (N=54; 9 post-study interviews) that compared five daily messaging approaches: random templates, contextual bandit with templates, LLM generation, hybrid bandit+LLM, and LLM with interaction history. LLM-based approaches were rated substantially more helpful than templates, but no significant differences emerged among LLM conditions. Unexpectedly, bandit optimisation for BCTs selection yielded no additional perceived helpfulness compared with LLM-only approaches. Unconstrained LLMs focused heavily on a single BCT, whereas bandit systems enforced systematic exploration-exploitation across techniques. Quantitative and qualitative findings suggest contextual acknowledgement of user input drove perceived helpfulness. We contribute design suggestions for reflective AI health behaviour change systems that address a trade-off between structured exploration and generative autonomy.

Problem

• Behaviour Change Techniques
• Personalised Health Interventions
• Contextual Bandits
• Large Language Models
• Structured Exploration

Innovation

• Contextual Bandits
• Large Language Models
• Behaviour Change Techniques
• Personalised Health Interventions
• Structured Exploration