An artifact-resilient multimodal bioelectronic patch for monitoring biochemical and motor states in Parkinson's disease.

Personalized management of Parkinson's disease (PD) requires continuous tracking of motor fluctuations and medication-associated biochemical dynamics during daily living. Here, we report the Biochemical and Inertial Multimodal Assessment Patch (BIMAP), a fully integrated multimodal bioelectronic patch that combines multiplexed sweat biosensing of levodopa (L-Dopa), glucose (Glu), and ascorbic acid (AA) with triaxial accelerometry for biochemical-kinematic monitoring of PD-related states. A microfluidic sampling interface is incorporated to stabilize sweat transport and maintain sensor-fluid contact during motion. We further develop a machine-learning-assisted Multimodal Adaptive Gating Network (MAG-Net) to adaptively weight biochemical and inertial features according to the instantaneous motion state. This strategy reduces the contribution of motion-perturbed electrochemical features, while retaining L-Dopa-related biochemical context for distinguishing medication-associated motor changes from frequency-matched daily activities. On-body feasibility tests involving healthy volunteers and participants with PD showed that the patch could track intake-associated biochemical changes and temporally aligned changes in estimated sweat L-Dopa and motor activity after medication administration. The multimodal signals further enabled exploratory discrimination of four protocol-defined medication-motion states under the tested conditions. These results support the feasibility of artifact-resilient biochemical-kinematic monitoring in wearable settings.