WEQA: Wearable hEalth Question Answering with Query-Adaptive Agentic Reasoning

WEQA introduces a query-adaptive agent framework that integrates LLM reasoning with wearable health data analysis, addressing the challenges of high-dimensional, continuous data from various sensor modalities. The framework utilizes an LLM controller to dynamically route queries to appropriate analytical tools and pretrained models, achieving a 24% accuracy improvement over existing baselines. This development is significant for practitioners as it enhances the effectiveness of medical question answering systems by leveraging specialized analytical capabilities tailored to wearable health data.