Enhancing Type 1 Diabetes Immunological Risk Prediction with Continuous Glucose Monitoring and Genetic Profiling.

Area of Science:

• Endocrinology
• Immunology
• Biomedical Engineering

Background:

• Early identification of individuals at high risk for type 1 diabetes (T1D) is crucial for intervention.
• Islet autoantibodies (AB), continuous glucose monitoring (CGM), and T1D genetic risk scores (GRS) are key indicators of early T1D risk.

Purpose of the Study:

• To develop and validate a machine learning model for classifying T1D risk based on AB levels.
• To predict early T1D risk by integrating CGM data and T1D GRS.

Main Methods:

• Utilized CGM data and T1D GRS from 39 AB-positive T1D relatives.
• Employed a linear support vector machine (SVM) with recursive feature elimination (RFE) for AB classification.
• Evaluated model performance using cross-validation and receiver operating characteristic (ROC) / precision-recall (PR) curves.

Main Results:

• The SVM model, using T1D GRS and incremental AUC, achieved high accuracy in classifying individuals with 1 versus ≥2 AB (AUC-ROC: 0.93, AUC-PR: 0.89).
• Significant differences in post-meal glucose levels and GRS were observed between AB groups (P=0.020 and P=0.001, respectively).

Conclusions:

• A machine learning approach combining CGM and T1D GRS effectively assesses T1D immunological risk.
• This method enables early risk assessment, facilitating timely interventions for individuals at high risk of T1D.