Machine-Learning-Assisted CRISPR/Cas12a Biosensors for Monitoring Organophosphorus Pesticide Degradation

Owing to the severe environmental and health issues posed by organophosphorus pesticides (OPs), a dual-enzyme cascade biosensing platform based on manganese dioxide (MnO2) and CRISPR/Cas12a was developed in this study. Smartphones were innovatively integrated with a stacked ensemble learning (SEL) model for ultrasensitive detection and dynamic monitoring of OPs in environmental water samples. A dual-enzyme cascade signal amplification strategy was used to construct this sensing platform. Acetylcholinesterase (AChE) catalyzes the generation of thiocholine (TCh), which, in turn, regulates the degradation of MnO2 nanosheets, releasing Cas12a activators and generating a fluorescence signal. Owing to the irreversible inhibition of AChE activity by OPs, dichlorvos (DDVP) was successfully detected, with a detection limit as low as 4.62 pg/mL. Additionally, the SEL model, integrated into the smartphone biosensing platform and incorporating random forest (RF), XGBoost, and ridge regression algorithms, exhibited strong performance in detecting OPs after optimization (R2 = 0.9985). In real water samples, the SEL model achieved a recovery rate of 93.1-103.1%, and the degradation kinetics of DDVP were successfully monitored over 24 h, revealing significant differences in DDVP degradation rates across various water matrices. This study is the first to report the integration of CRISPR/Cas12a biosensing technology with an SEL model-driven smartphone detection platform, providing a novel approach for sensitive, portable, and intelligent monitoring of OPs and offering new insights for water quality monitoring and early detection of environmental risks.