Development Of Predictive Model for Diabetes Medication Adherence Using Algorithms.

Abstract

Type 2 diabetes (T2DM) is a persistent metabolic disorder characterized by its complex interaction with both environmental factors and genetic predisposition. It poses a significant global health challenge, steadily increasing in prevalence and presenting substantial difficulties for healthcare systems and individuals alike. Managing T2DM effectively requires a comprehensive approach involving lifestyle changes, medications, and sometimes insulin therapy to regulate blood sugar levels and mitigate associated complications. In Pakistan, the prevalence of Type 2 Diabetes Mellitus (T2DM) has risen to concerning levels, posing a formidable health issue nationwide. Our study centers on developing a web-based application utilizing a decision tree regressor to forecast patients' HbA1c levels and medication adherence. Validation of the application includes analyzing gene expression of GLUT4. Additionally, association studies involving expression are conducted to potentially integrate this markers into future models. The model achieved an accuracy of 80% with metrics showing a mean squared error of 0.143, mean absolute error of 0.15, and an R2 value of 0.88. Future studies could explore incorporating GLUT4 expressions to enhance predictive accuracy further.