(697c) Machine Learning Assisted Development of Electrochemical Cortisol Sensor Based on Electropolymerized Molecularly Imprinted Polymer

Molecularly imprinted polymers (MIPs) with tailored biomolecular recognition hold great promise to substitute antibodies used in biosensors and bioassays, which have clear advantages of low cost, easy fabrication, good stability, excellent durability, and long lifetime. Electropolymerization is a fast and facile method to directly synthesize MIP sensing elements in-situ on the working electrode, enabling ultralow-cost and easy-to-manufacture electrochemical biosensors. However, due to the high dimensional design space of electropolymerized MIPs (e-MIPs), the development of e-MIPs is challenging and lengthy mainly based on trial and error without proper guidelines. Along with the advancement of computational power, machine learning techniques have become the mainstream to facilitate various engineering applications through data-driven surrogate modeling (or metamodeling). We herein demonstrate a case study on the development of an electrochemical cortisol sensor based on e-MIPs assisted by an integrated data-driven framework established upon the small-sized experimental data.

Cortisol, popularly called “stress hormone”, is a highly valuable biomarker to be measured for stress management and personalized health monitoring. Pyrrole is selected as the functional monomer, and cortisol-imprinted-polypyrrole-based sensors are fabricated with 72 sets of synthesis parameters with replicates, including pyrrole concentration, cortisol concentration, number of electropolymerization CV cycles, electropolymerization CV scan rate, number overoxidation cycles for template elution. Their sensing performances are measured using a 12-channel potentiostat to construct the subsequent data-driven framework, which greatly improves experimental efficiency and ensures sensor reproducibility and data quality. The Gaussian process (GP) is employed as the mainstay of the integrated framework, which enables the probabilistic decision-making that accounts for various uncertainties in the synthesis and measurements. The Sobol index-based global sensitivity is then performed upon the GP surrogate model to elucidate the impact of e-MIPs’ synthesis parameters on sensing performance and interrelations among parameters. Based on the prediction of the established GP model and local sensitivity analysis, synthesis parameters are optimized and validated by experiment, which leads to remarkable sensing performance enhancement (1.5-fold increase in sensitivity). The proposed framework is novel in biosensor development, which is expandable and also generally applicable to other sensing materials development.