Energy-Decoupled Photoelectrochemical and Pressure Dual-Mode Biosensors for Machine-Learning-Assisted Immunoassay.

Reliable and ultrasensitive detection of cardiac troponin I (cTnI) remains challenging due to the signal interference and limited self-validation capability of conventional biosensors. Herein, we report an orthogonal dual-mode biosensing strategy that integrates photoelectrochemical (PEC) sensing with catalysis-induced pressure transduction (CIPT) to achieve energy-decoupled signal generation and data-level cross-validation. Specifically, a multifunctional plasmonic Z-scheme probe, CdS@CdIn2S4/AuPt, was first synthesized to trigger in situ assembly of ZnO-based dual Z-scheme heterostructures upon cTnI recognition. This configuration simultaneously amplifies the photocurrent via enhanced charge separation and catalyzes the hydrolysis of ammonia borane to produce a macroscopic pressure change. These two outputs originate from distinct energy conversion pathways─interfacial charge transfer versus bulk gas expansion─thereby establishing an energy-decoupled dual-signal system. 3D-printed portable devices were further fabricated for signal acquisition, and a machine learning (ML)-assisted data analysis was employed for multivariate data fusion. The smart sensing platform exhibits high sensitivity for cTnI with limits of detection of 41.94 fg·mL-1 (PEC) and 164.80 fg·mL-1 (CIPT). The partial least-squares (PLS) model enables cross-channel validation, achieving a high prediction correlation (R2 > 0.9972) and improving recovery rates (98.74%-102.69%) in human serum. This work offers a robust and reliable pathway for the construction of energy-decoupled sensing platform for point-of care testing applications.