Proof-of-Concept Study of Using Supervised Machine Learning Algorithms to Predict Self-Care and Glycemic Control in Type 1 Diabetes Patients on Insulin Pump Therapy
Authors' conclusion
These proof-of-concept analyses demonstrate the feasibility of using SMLAs to develop clinically relevant predictive models of adherence with IPSMB criteria and glycemic control within 6 months. Subject to further study, nonlinear prediction models may perform better.
Study context
- Population
- Adults with T1D on insulin pump therapy (≥6 months), single centre
- Participants
- 100 adults
- Exposure
- Three supervised machine learning algorithms: Logistic regression (LR) vs random forest (RF) vs K-nearest neighbor (k-NN)
- Outcome
- Prediction of meeting insulin pump self-management behavioral (IPSMB) criteria and achieving good glycemic response; discrimination (AUC-ROC) and calibration (Brier score)
- Follow-up
- Within 6 months
Methodological detailssingle-center, retrospective chart review, proof-of-concept, repeated three-fold cross-validation
- Single-center
- Retrospective chart review
- Proof-of-concept
- Repeated three-fold cross-validation
Main finding
As reported in the abstract
Adherence with IPSMB criteria: discrimination (AUC) and calibration
| Group | Value | AUC | Brier |
|---|---|---|---|
| LR | 0.74 | 0.74 | |
| RF | 0.74 | 0.74 | 0.151 |
| k-NN | 0.72 | 0.72 |
Value
LR0.74
RF0.74
k-NN0.72
AUC
LR0.74
RF0.74
k-NN0.72
Brier
LR— not stated in the abstract
RF0.151
k-NN— not stated in the abstract
Observational data: associations, not effects.
Further findings
| Outcome | LR | RF | k-NN |
|---|---|---|---|
| Good glycemic response: discrimination (AUC) and calibration · within 6 months | |||
| AUC | 0.81 | 0.80 | 0.78 |
| Brier | — not stated in the abstract | 0.099 | — not stated in the abstract |
— not stated in the abstract
Good glycemic response: discrimination (AUC) and calibration · within 6 months
- Predictors of good glycemic response: baseline HbA1c, entering carbohydrates, following recommended bolus dose
- Comparable discriminatory power; RF model better calibrated
Original publication
Proof-of-Concept Study of Using Supervised Machine Learning Algorithms to Predict Self-Care and Glycemic Control in Type 1 Diabetes Patients on Insulin Pump Therapy. Endocr Pract. Published 8 March 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →