Development and validation of a machine learning-based predictive model for carotid plaque in type 2 diabetes.

Area of Science:

• Cardiovascular Medicine
• Medical Informatics
• Diabetes Research

Background:

• Carotid plaque is a key indicator of atherosclerotic burden and cardiovascular disease risk.
• Patients with type 2 diabetes mellitus (T2DM) have a higher risk of developing carotid plaques.
• Predictive models for carotid plaque in T2DM patients are crucial for early intervention.

Purpose of the Study:

• To develop and validate a machine learning-based predictive model for carotid plaque in patients with type 2 diabetes mellitus.
• To identify key clinical variables contributing to carotid plaque formation in this high-risk population.

Main Methods:

• Utilized data from 2,288 patients, with 1,716 diagnosed with carotid plaques via ultrasound.
• Applied seven machine learning algorithms to build the prediction model.
• Employed Shapley Additive Explanations (SHAP) for feature importance and evaluated model performance using AUC-ROC, sensitivity, and specificity.

Main Results:

• The logistic regression model demonstrated superior performance in discrimination, calibration, and clinical utility.
• Key predictors identified include age, BMI, glycated hemoglobin, hypertension history, monocyte count, neutrophil percentage, RBC count, sex, eGFR, and statin use.
• Confirmed that machine learning models can effectively predict carotid plaque risk using conventional clinical data.

Conclusions:

• The developed predictive model aids primary care providers in identifying T2DM patients at high risk for carotid atherosclerotic plaques.
• Enables targeted preventive strategies for cardiovascular events in this population.
• Facilitates improved clinical outcomes through early detection and management.