Ultraflexible Multimodal Sensors with Stable Physical Sensing and Disposable Electrochemical Calibration for Long-Term Glucose Prediction.

Wearable electrochemical biosensors offer direct biochemical readouts but often suffer from enzyme deactivation, biofluid fluctuation, fouling, and calibration drift during prolonged on-body use. Here, we report an ultrathin (∼1.3 μm thick) multimodal sensing system that decouples whole-day physical monitoring from short-term disposable electrochemical calibration. The system integrates stable physical sensors for electrocardiogram, pulse, and temperature acquisition with a detachable electrochemical glucose module that provides personalized calibration labels during initial use. At the material level, the pulse sensor combines an embedded Ag nanowire network with a cracked Cr/MoO3/Cr/Au layer in an ultrathin polyimide matrix, enabling both strain sensitivity and cyclic stability. The resulting strain sensor shows a gauge factor of 5.49 in the 0-4% strain range, a response time of 55 ms, and durability over 15,000 stretching cycles. The detachable glucose sensor employs a co-crosslinked porous enzymatic matrix to delay active-site saturation, achieving a broad linear range of 0-20 mM. After three days of personalized calibration, a random-forest model using 60 s recordings of electrocardiogram, pulse, and temperature signals yields an overall tolerance-based accuracy of 81.36% and a mean absolute relative difference (MARD) of 9.12%. Furthermore, in individual on-body tracking over one day and 15 days, the glucose predictions achieved an accuracy of at least 80%. This stable-physical-sensing/disposable-electrochemical-calibration architecture provides a practical strategy for personalized glucose prediction while mitigating the long-term instability of wearable electrochemical biosensors.