P W C Prasad, Daniel Patricko Hutabarat, Md Shohel Sayeed, Golam Md Mohiuddin
The synthesis identified wearable and multimodal sensing, edge cloud architectures, AI-driven behavioural modelling, and just-in-time adaptive interventions (JITAIs) as recurring technological and intervention components. Comparatively stronger empirical support was observed for continuous monitoring, behavioural and physiological pattern recognition, remote patient monitoring, elderly care, and selected chronic disease applications. In contrast, evidence for advanced adaptive coaching, digital twins, metaverse-enabled healthcare, behavioural simulation, and emerging AI or large language model (LLM) enabled systems was more heterogeneous and frequently derived from prototype, simulation-based, or early-stage evaluations. Across the evidence base, substantial variation was observed in study designs, populations, datasets, intervention types, validation settings, and outcome measures, limiting direct cross-study comparison and conclusions regarding comparative effectiveness. Key limitations included insufficient longitudinal and clinical validation, incomplete reproducibility, fragmented interoperability, heterogeneous evaluation metrics, and unresolved privacy and governance challenges.
INTRODUCTION: The convergence of the Internet of Things (IoT), artificial intelligence (AI), and behavioural analytics has contributed to the emergence of the Internet of Behaviours (IoB) as a framework for personalized and context-aware digital health interventions. This study systematically reviews and synthesizes the architectures, sensing technologies, behavioural analytics methods, machine learning approaches, intervention strategies, and evaluation practices underlying IoB-enabled adaptive health coaching systems.
METHODS: Following the PRISMA 2020 guidelines, a systematic literature review was conducted across IEEE Xplore, Scopus, Web of Science, PubMed, and ACM Digital Library. A total of 75 eligible peer-reviewed studies published between 2015 and 2025 met the predefined eligibility criteria and were included in the final evidence synthesis.
RESULTS: The synthesis identified wearable and multimodal sensing, edge cloud architectures, AI-driven behavioural modelling, and just-in-time adaptive interventions (JITAIs) as recurring technological and intervention components. Comparatively stronger empirical support was observed for continuous monitoring, behavioural and physiological pattern recognition, remote patient monitoring, elderly care, and selected chronic disease applications. In contrast, evidence for advanced adaptive coaching, digital twins, metaverse-enabled healthcare, behavioural simulation, and emerging AI or large language model (LLM) enabled systems was more heterogeneous and frequently derived from prototype, simulation-based, or early-stage evaluations. Across the evidence base, substantial variation was observed in study designs, populations, datasets, intervention types, validation settings, and outcome measures, limiting direct cross-study comparison and conclusions regarding comparative effectiveness. Key limitations included insufficient longitudinal and clinical validation, incomplete reproducibility, fragmented interoperability, heterogeneous evaluation metrics, and unresolved privacy and governance challenges.
DISCUSSION: Based on recurring architectural, technological, and methodological patterns identified across the included studies, this review proposes a synthesized reference architecture and evaluation framework for IoB enabled adaptive health coaching. These evidence-informed conceptual synthesis frameworks are intended to guide future research and implementation rather than serve as universally validated standards. Future research should prioritize longitudinal and multi-site validation, representative datasets, reproducible reporting, interoperable clinical integration, privacy-preserving analytics, and the combined evaluation of technical, behavioural, and clinically meaningful outcomes.