Uncovering Behavioral Themes in Mobile Health Counseling: A Topic Modeling Study of Public Health Center Data

Abstract

This study examines linguistic and behavioral patterns embedded in mobile healthcare counseling messages by applying text-mining techniques to a large set of public health communication data. A total of 524 counseling sessions collected from the Korean Public Health Center Mobile Healthcare Program between 2021 and 2023 were analyzed. After comprehensive preprocessing and topic modeling using Latent Dirichlet Allocation (LDA), five salient behavioral themes were identified: exercise and motivation, dietary management, stress and sleep, adherence and monitoring, and clinical risk awareness. These topics collectively highlight that lifestyle modification—rather than disease-specific instruction—is the central focus of digital counseling within public healthcare settings. The findings reveal that exercise and dietary guidance dominated the counseling narratives, while emotional expressions such as encouragement, reassurance, and motivational feedback were strongly associated with higher user engagement. These patterns underscore the crucial role of affective communication in sustaining participation and supporting behavior change in mobile health environments. The analysis further demonstrates that consistent monitoring cues and feedback loops function as important mechanisms that reinforce adherence. Methodologically, this study shows how text-mining approaches can effectively quantify behavioral tendencies and latent themes within unstructured counseling data, offering an analytical framework that complements traditional clinical and physiological metrics. Practically, the extracted behavioral insights provide valuable implications for designing AI-assisted digital coaching systems capable of delivering tailored, empathetic, and context-aware guidance. Overall, this study contributes to the growing body of research on digital health communication by presenting a scalable approach to understanding public health counseling language. The findings support future interdisciplinary efforts to integrate linguistic analytics into preventive healthcare, automated coaching platforms, and personalized intervention strategies.