Machine Learning-based QSAR Modeling of α-Amylase Inhibitors as Potential Antidiabetic Agents

Modulating the enzymatic activity of amylase holds significant therapeutic promising in diabetes mellitus, primarily due to its ability to catalyze the hydrolysis of starch into simpler sugars. This study employs computational models utilizing experimental datasets, focusing on designing inhibitors of α-amylase. Despite limited information regarding in silico predictive models’ capability related to α-amylase, we collected various data and applied multiple linear regression-based machine learning technique (MLR-ML) to forecast the inhibitory activity of α-amylase inhibitors as antidiabetic agents. The model was developed using a dataset comprising compounds relevant to α-amylase's preventive action and the model exhibited R2 correlation values of 0.887 and 0.887 for training and prediction sets, respectively. These findings underscore the efficacy of an in silico approach employing machine learning (ML) techniques in identifying potential antidiabetic compounds. Collectively, our study demonstrates that this approach is a viable strategy for regulating postprandial hyperglycemia and mitigating diabetes risk.