Rule-Based Versus Large Language Model Approaches for Creating mHealth Interventions for Loneliness and Social Isolation

Abstract

Loneliness and social isolation are pervasive public health challenges that require innovative and scalable solutions. Mobile health (mHealth) interventions offer a promising way to address these issues. However, while traditional rule-based systems have been widely used, the potential benefits of integrating large language models (LLMs) into mHealth interventions have not yet been thoroughly explored. This study had two main objectives. The first was to design and create two functionally aligned mHealth prototypes: one based on a rule-based system and the other driven by an LLM, both aimed at alleviating loneliness and social isolation. The second was to evaluate these prototypes in terms of user experience and system performance. The study examined the key trade-offs and synergies between LLM-driven and rule-based systems. The findings provide actionable insights for optimizing mHealth interventions that target loneliness and social isolation. These outcomes contribute to evidence-based frameworks and establish a foundation for future longitudinal trials.