Multilevel Integration of Artificial Intelligence and Wearable Biosensors for Precision Health and Clinical Translation

Abstract Wearable biosensors are evolving from passive data loggers toward proactive tools for health monitoring, playing an increasingly important role in personalized and decentralized healthcare. Artificial intelligence (AI) helps address key limitations in this field, including constrained sensing accuracy, complex signal interpretation, and challenges in multimodal data integration. This review summarizes a three-level integration framework, in which AI supports primary-level integration by interpreting acquired signals, secondary-level integration by co-optimizing the sensing process, and tertiary-level integration by producing system-level health judgments. From a technical perspective, AI is progressing from a data-processing tool to a co-design framework that jointly optimizes materials, devices, and signal processing strategies, thereby improving overall performance and stability. In application scenarios, multimodal data fusion and intelligent modeling support decision pipelines that are oriented toward closed-loop operation, enabling more accurate prediction and management of stress, metabolic disorders, neurological diseases, and ophthalmic conditions. These advances are gradually shifting healthcare paradigms from diagnosis-oriented approaches toward earlier and more proactive intervention. Nevertheless, challenges remain in data standardization, model interpretability, and the trade-off between system integration and energy consumption. Future progress is likely to be driven by generative AI, physics-informed modeling, personalized digital twins, and decentralized health systems, collectively pushing wearable biosensors toward higher levels of autonomy and clinical translation.