The rising prevalence of diabetes has made high-precision, continuous glucose monitoring a central challenge in intelligent health management. This review systematically examines recent advances in artificial intelligence (AI)-enabled electrochemical glucose sensing. We overview mainstream machine learning (ML) algorithms and their applications in electrochemical signal modeling, denoising, feature extraction, and adaptive calibration. Four representative platform categories are analyzed: enzyme-based sensors, non-enzymatic and affinity-based systems, electrochemiluminescence (ECL) sensing, and wearable continuous monitoring platforms. Representative studies show that ML algorithms such as support vector regression (SVR), XGBoost, and random forest (RF) can substantially improve analytical performance, with detection limits reduced from 100 nM to 10 nM and R2 values exceeding 0.9 in selected cases. The deep integration of AI and electrochemical sensing is transforming glucose detection from passive measurement to active cognition, namely to sensing systems that not only record signals but also perform adaptive denoising, drift correction, individualized calibration, and predictive decision support. This transition points toward explainable, low-power, privacy-aware, and wearable personalized health monitoring systems.
Qi et al. (Thu,) studied this question.