Risk-Profile and Feature Selection Comparison in Diabetic Retinopathy

Area of Science:

• Medical Informatics
• Ophthalmology
• Endocrinology

Background:

• Diabetic retinopathy is a major microvascular complication affecting 27.50% of type 2 diabetes patients in Mexico.
• Early identification of risk factors is crucial for managing diabetic retinopathy.

Purpose of the Study:

• To develop a predictive model for identifying diabetic retinopathy risk factors in the Mexican population.
• To utilize machine learning for enhanced diagnostic capabilities.

Main Methods:

• Employed machine learning techniques, including Boruta for feature selection and random forest for classification.
• Utilized a dataset of 298 subjects with clinical and paraclinical features.
• Model evaluation included sensitivity, specificity, Area Under the Curve (AUC), and Receiver Operating Characteristic (ROC) curve analysis.

Main Results:

• The predictive model achieved an AUC of 69%.
• Identified key risk factors: creatinine, lipid treatment, glomerular filtration rate, waist hip ratio, total cholesterol, and high-density lipoprotein.
• High cholesterol levels were associated with a 3.5916 times increased odds ratio for diabetic retinopathy.

Conclusions:

• The developed methodology serves as a preliminary computer-aided diagnosis tool.
• It assists clinicians in identifying diabetic retinopathy (DR) and making informed decisions.
• Highlights the importance of monitoring specific biomarkers for diabetic retinopathy risk assessment.