Predicting stimulated C-peptide in type 1 diabetes using machine learning: a web-based tool from the T1D exchange registry
Authors' conclusion
This ML model provides a practical, non-invasive tool for estimating beta-cell function in T1D and is available online at https://cpeptide.streamlit.app.
Study context
- Population
- Individuals with T1D in the T1D Exchange Registry with complete mixed-meal tolerance test (MMTT) and clinical information; predictors: age at diagnosis, diabetes duration, HbA1c, non-fasting glucose, non-fasting C-peptide
- Participants
- 319 individuals
- Comparison
- Four machine learning (ML) algorithms predicting MMTT-stimulated C-peptide categories
- Outcome
- Prediction of MMTT-stimulated C-peptide categories from routine clinical data
Methodological detailsmachine learning model development and validation, random 70%/30% training/test split, 10-fold cross-validation, registry-based
- Machine learning model development and validation
- Random 70%/30% training/test split
- 10-fold cross-validation
- Registry-based
Performance
As reported in the abstract
Performance of best model (RF), cross-validation
| Test or model | AUC | Sensitivity | Specificity |
|---|---|---|---|
| RF | 0.94 (95% CI: 0.92-0.96) | 0.84 (95% CI: 0.80-0.89) | 0.92 (95% CI: 0.90-0.94) |
RF
AUC0.94 (95% CI: 0.92-0.96)
Sensitivity0.84 (95% CI: 0.80-0.89)
Specificity0.92 (95% CI: 0.90-0.94)
Further findings
Model performance in the test set
- Test-set AUC
- 0.97
- Test-set sensitivity
- 88%
- Test-set specificity
- 94%
Measurable C-peptide after MMTT among individuals with undetectable non-fasting C-peptide
- Undetectable non-fasting C-peptide but measurable levels after MMTT
- 17.7%
Original publication
Predicting stimulated C-peptide in type 1 diabetes using machine learning: a web-based tool from the T1D exchange registry. Diabetes Res Clin Pract. Published 4 September 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →