Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{199638,
        author = {Astha Raut and Gunakshi Gujar and Samiksha Dhomane and Aditi Raut},
        title = {Risk Prediction of Type-1 Diabetes Using Machine Learning},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {14715-14720},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=199638},
        abstract = {Diabetes mellitus affects over 537 million adults worldwide, but clinical deterioration is usually detected only after irreversible metabolic damage has happened. This paper introduces RiskEngine, a complete machine learning pipeline that combines Continuous Glucose Monitoring (CGM) telemetry, HbA1c biochemistry, patient demographics, and ICD-10 diagnostic codes to create personalized 90-day deterioration risk scores. Raw 15-minute CGM readings are compiled into daily clinical metrics such as Time-in-Range (TIR 70–180 mg/dL), mean glucose, glycaemic variability, and the burden of hyperglycaemia and hypoglycaemia. It also builds multi-horizon rolling features over 30-, 90-, and 180-day periods, including a linear regression glucose slope. Deterioration labels follow a dual criterion: future hyperglycaemia burden exceeding 35% or an HbA1c increase of 0.5% or more within 90 days. An XGBoost classifier with isotonic probability calibration scores an AUROC of 0.921, an AUPRC of 0.882, and a Brier Score of 0.091 on a patient-stratified held-out test set, surpassing Logistic Regression (0.843), SVM (0.851), Random Forest (0.876), and LSTM-CGM models (0.891). SHAP-based explainability points to glucose slope as the top predictor. A Streamlit clinical dashboard implements all components with a 4-tab interface that includes calibration diagnostics and Decision Curve Analysis.},
        keywords = {Continuous glucose monitoring, diabetes risk prediction, XGBoost, SHAP explainability, isotonic calibration, time-in-range, HbA1c, decision curve analysis, clinical decision support, rolling window features.},
        month = {April},
        }