Comparing Three Data Mining Algorithms for Identifying the Associated Risk Factors of Type 2 Diabetes

Area of Science:

• Medical Informatics
• Biostatistics
• Public Health

Background:

• Type 2 diabetes presents a significant global health challenge.
• Identifying risk factors is crucial for effective management and prevention.
• Health service providers and administrators require robust predictive models.

Purpose of the Study:

• To develop and compare statistical models for identifying type 2 diabetes risk factors.
• To evaluate the performance of artificial neural network (ANN), support vector machines (SVM), and multiple logistic regression (MLR) models.
• To assess model efficacy using demographic, anthropometric, and biochemical data.

Main Methods:

• Applied ANN, SVM, and MLR models to a dataset of 9528 individuals from Mashhad, Iran.
• Utilized demographic, anthropometric, and biochemical characteristics for model training and testing.
• Compared model performance using Receiver Operating Characteristic (ROC) curves and key performance metrics.

Main Results:

• The prevalence of type 2 diabetes in the study population was 14%.
• ANN achieved 78.7% accuracy, 63.1% sensitivity, and 81.2% specificity.
• SVM and MLR showed comparable, yet lower, performance metrics, with ROC AUC values of 0.73 and 0.70 respectively, versus 0.71 for ANN.

Conclusions:

• Artificial neural network (ANN) models exhibit superior performance over SVM and MLR for identifying type 2 diabetes risk factors.
• ANN models can be effectively utilized for early detection and risk stratification of type 2 diabetes.
• The study highlights the potential of advanced statistical modeling in public health initiatives for diabetes.