Experimental results demonstrate the method outperforms commercial general and medical LLMs and open-source models and human evaluation further supports the quality of the generated guidance and the feasibility of personalized sleep guidance with SLMs.
Abstract
Sleep monitoring using wearable data has shown promise for personal health, yet large language model (LLM)-based summarization and question answering remain insufficient for personalized sleep guidance. Training specialized models, however, often requires costly expert annotation. Moreover, privacy and accessibility concerns motivate lightweight, local deployment for end users. We present a two-stage framework to address these challenges. Specifically, in Stage~1, a multi-agent LLM pipeline reasons structured sleep guidance from unannotated wearable records, enabling scalable dataset construction. Stage~2 distills guidance reasoning trajectories into small language models (SLMs) through supervised fine-tuning and integrates a training-free Best-of-$N$ selection strategy to enhance inference. Experimental results demonstrate our method outperforms commercial general and medical LLMs and open-source models. Human evaluation further supports the quality of the generated guidance and the feasibility of personalized sleep guidance with SLMs.
Accurate sleep staging is essential for both clinical diagnosis and long-term sleep monitoring. However, most existing deep learning approaches rely on large-scale architectures and server-based inference, which limits their applicability in mobile and privacy-sensitive scenarios. This study evaluates whether clinicall...
Dongdong Zhou, Zhaowen Wang, Jiacheng Zhang et al.· IEEE journal of biomedical a...· 0 citations
A Concept-Integrated Transformer with LLM-guided concept supervision for explainable prediction from mobile sensing data that shows the clearest difference between high and low negative affect groups in AFFECT and ties for the highest F1 score on AFFECT.
Yu-Ning Wang, Iman Azimi, Amir M. Rahmani et al.· 0 citations
BALMS is the first systematic benchmark of LLM-based agentic systems for longitudinal mental health sensing and highlights the need for longitudinal mental health agents that selectively retrieve history, ground temporal evidence, and reason over interpretable behavioral features.
Y. Wu, Arvind Pillai, Yu-Liang Chen et al.· 0 citations
Findings support LUMINA as an exploratory foundation for personalized and context-aware wellbeing monitoring, and incorporate AI-driven components for sleep-related prediction and emotional analysis as auxiliary signals for DT updating.
Domenico Rossi, L. Fiore, Alessia Auriemma Citarella et al.· Companion Publication of the...· 0 citations
Wearable health questions often combine data retrieval, longitudinal analysis, and health advice over structured records. Prompting a single large language model with a complete record and a composite query obscures whether every request is executed and which evidence supports the answer. We propose a task-oriented mul...
This work introduces HealthLoopQA, a comprehensive diagnostic benchmark for evaluating LLM reasoning over continuous diabetes monitoring data, and identifies a broader phenomenon of In-context Laziness under long-context prompting, highlighting critical open challenges in deploying LLMs for rigorous long-horizon medica...
Yu-Chen Niu, Ya-Nan Ma, Srinivasan Nandakumar et al.· 0 citations