Mechanism-Driven Self-Powered Biosensing: Integrating Entropy-Controlled Nanocatalysis with Machine Learning on a Generalizable Hydrogel Platform

Conventional self-powered biosensors often face a trade-off between sensitivity and operational stability, largely hindered by the intrinsic instability of biocatalysts and linear signal-transduction mechanisms. Herein, we propose a mechanism-driven sensing paradigm that synergistically integrates entropy-driven DNA nanotechnology, ultrasmall PtNPs, and machine learning on a solid-state hydrogel electrolyte. The target-triggered entropy-driven circuit acts as a precise molecular switch, programmably releasing ultrasmall PtNPs (2.0–4.5 nm) to modulate the oxygen reduction reaction (ORR) kinetics via a two-electron pathway. This catalytic process is powered by an air-breathing zinc-air fuel cell featuring a TiCN-doped hydrogel electrolyte, which ensures rapid ion migration and robust performance across a broad temperature range (−20–25 °C). Crucially, the sensor exhibits a complex, nonlinear response spanning 10 orders of magnitude (10–15 to 10–6 M), which defies conventional calibration models. To address this, a Ridge regression machine-learning algorithm is employed to decode the signal-concentration relationship, achieving high-fidelity prediction with an R2 of 0.9914. The platform demonstrates exceptional universality, enabling ultrasensitive detection of biomolecular targets in complex matrices with a 263-fold miniaturization compared to enzymatic counterparts. This work establishes a generalizable and intelligent framework for next-generation self-powered diagnostics, bridging molecular engineering, electrocatalysis, and computational intelligence.