Chenshu Liu, Haolin Fan, Minwoo Kim, Tong Zhou, Pinyi Yang, Lingdi Zhao, Y. M. Wang, Ziyuan Che, Chia‐Wei Liu, Bingbing Li, Yangzhi Zhu
Recent advances in multimodal wearable biosensing enable continuous, noninvasive or minimally invasive monitoring of physical, physiological, and biochemical states in daily life. In parallel, multidomain AI architectures are increasingly capable of fusing heterogeneous streams, creating new opportunities for scalable, patient-specific health analytics. Yet many sensor-AI systems remain narrow, tracking limited parameters, and often emphasize real-time signals while underutilizing longitudinal clinical context and structured medical knowledge that could strengthen clinical reasoning. Here, we propose a pathway to decentralized healthcare that unites multimodal wearable biosensing with multidomain AI. We review recent progress across wearable sensing modalities and summarize how multisensory fusion can improve patient profiling, enhance diagnostic discrimination, and enable earlier risk prediction. We then describe AI pipelines that integrate biosensor measurements with electronic health records and curated medical literature and knowledge graphs to support evidence-grounded decision support. Finally, we discuss remaining challenges, including data quality and cross-modality alignment, privacy and governance for cross-domain data sharing, and robust generalization under real-world heterogeneity. We highlight emerging opportunities in continual learning, retrieval-augmented reasoning, and closed-loop therapeutics. This "from biosignals to decisions" framework advances AI-enabled decentralized healthcare by shifting actionable insights from the clinic into everyday environments.