Abstract. As a global chronic disease, diabetes has a serious impact on human health and imposes a significant economic burden. Facing this challenge, researchers are actively developing and optimizing predictive models to improve early diagnosis and management of diabetes. This paper analyzed the influence of independent variables on the risk of diabetes by logistic regression from 8 aspects including body mass index (BMI), glycosylated hemoglobin (HbA1c) and heart disease. The logistic regression model, an effective binary classification method, optimizes parameters using maximum likelihood estimation to predict diabetes probability. The model will be evaluated by ROC curve, cross-validation, standardized residual analysis and confusion matrix to comprehensively test its predictive power, stability and classification performance. The results showed that hemoglobin A1c level (HbA1c.level) had the most significant effect on diabetes risk. Other relevant variables, including blood glucose level and body mass index (BMI), demonstrated significant positive correlations, particularly with hypertension and heart disease. The findings will enhance early diabetes identification and provide data to support the development of targeted prevention and intervention measures to reduce the burden of diabetes on individuals and society.
Read full abstract