Real-Time Monitoring of Midazolam Using Photoelectric Sensing for Safe Sedation Management in Critical Care

Midazolam sedation in Intensive Care Units (ICUs) requires precise monitoring to balance efficacy against life-threatening adverse events, yet conventional methods lack real-time capability and personalized risk assessment. This study pioneers a dual-modality platform integrating Bi–Bi 4 Ti 3 O[Formula: see text] nanoflower-based photoelectrochemical (PEC) sensor with interpretable Machine Learning (ML) to address this challenge. The optimized 20% Bi/Bi 4 Ti 3 O[Formula: see text] nanoflower heterostructure, engineered via controlled Bi doping, exhibits 4.5-fold enhanced photocurrent density (1.88[Formula: see text][Formula: see text]A/cm[Formula: see text] under visible light, enabled by surface plasmon resonance and Schottky barrier-mediated charge separation. The PEC sensor achieves a linear response to midazolam ([Formula: see text]) and exceptional stability. Mechanistic studies reveal midazolam acts as a hole scavenger, amplifying photocurrent proportional to concentration via adsorption-driven charge transfer. Complementarily, ML analysis identifies dose, neurological status and multiorgan dysfunction as key predictors, with the Random Forest model achieving 0.939 AUC and 94.6% specificity for adverse event prediction. The platform bridges nanomaterial innovation with explainable AI, demonstrating clinical-grade linear response ([Formula: see text]) and probabilistically calibrated risk stratification to mitigate the 25.9% adverse event rate in critical care. This dual-platform approach synergizes nanomaterial-enabled PEC sensing with interpretable ML, offering a novel strategy for personalized sedation management in critical care.