Using a robust model to detect the association between anthropometric factors and T2DM: machine learning approaches.

Area of Science:

• Endocrinology and Metabolism
• Biostatistics
• Medical Informatics

Background:

• Type 2 Diabetes Mellitus (T2DM) is a significant global health concern.
• Identifying reliable anthropometric predictors for T2DM is crucial for early detection and intervention.
• Previous models for T2DM prediction have varying degrees of success.

Purpose of the Study:

• To evaluate different predictive models for identifying key anthropometric factors linked to T2DM.
• To determine the most influential anthropometric measurements associated with T2DM.
• To compare the performance of K-nearest neighbor (KNN) with Artificial Neural Network (ANN) and Support Vector Machine (SVM) models.

Main Methods:

• Utilized a dataset of 9354 participants from the MASHAD study (ages 35-65).
• Analyzed 10 anthropometric factors and age, employing a K-nearest neighbor (KNN) model.
• Evaluated model performance using accuracy, sensitivity, specificity, precision, F1-score, ROC curves, and feature importance.

Main Results:

• Six anthropometric factors (MAC, WC, BRI, BAI, BMI, age) were identified as significant predictors.
• Body Roundness Index (BRI), Body Adiposity Index (BAI), and Mid-arm Circumference (MAC) were most associated in males.
• Body Mass Index (BMI), BRI, and MAC were most associated in females.
• The KNN model achieved approximately 93% accuracy with an Area Under the ROC Curve (AUC) of 0.985 (men) and 0.986 (women).

Conclusions:

• The KNN model accurately predicts T2DM association with anthropometric factors (93% accuracy).
• Optimizing the K parameter is vital for minimizing error rates in KNN models.
• Feature selection enhances KNN model efficiency and predictive accuracy for T2DM.