Development and external validation of an algorithm for self‐identification of risk for microvascular complications in patients with type 1 diabetes
Authors' conclusion
ML models, based on self-reported data, have the potential to serve as a self-identification tool, empowering T1D patients to understand their risks outside of hospital settings and encourage early engagement with healthcare services.
Study context
- Population
- Patients with T1D; models built on 5 of 15 patient self-reported variables
- Participants
- 911 vs 157
- Comparison
- Development set vs External validation set
- Development setn = 911
- External validation setn = 157
- Outcome
- AUROC of machine learning (ML) models for self-identification of DR, DN and DPN risk
Methodological detailsmachine learning (XGBoost), cross-validation, external validation via online survey, SHapley Additive exPlanation used for model interpretation
- Machine learning (XGBoost)
- Cross-validation
- External validation via online survey
- SHapley Additive exPlanation used for model interpretation
Performance
As reported in the abstract
AUROC of ML models for microvascular complications
- Internal validation AUROC, diabetic retinopathy (DR)
- 0.889
- Internal validation AUROC, diabetic nephropathy (DN)
- 0.844
- Internal validation AUROC, diabetic peripheral neuropathy (DPN)
- 0.839
- External validation AUROC, DR
- 0.762
- External validation AUROC, DN
- 0.718
- External validation AUROC, DPN
- 0.721
Further findings
Prevalence of DR
P = 0.013
Prevalence of DN
P = 0.162
Prevalence of DPN
P < 0.001
Original publication
Development and external validation of an algorithm for self‐identification of risk for microvascular complications in patients with type 1 diabetes. Diabetes Obes Metab. Published 25 November 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →