Identifying Predictors of Early Treatment Intensification in Individuals With Type 2 Diabetes Treated With GLP ‐1 Receptor Agonists: A Machine Learning–Based Analysis of a Large Real World Cohort
Authors' conclusion
Explainable AI identified a reproducible phenotype of patients associated with early intensification after GLP-1 RA initiation. These models may support earlier, more personalised treatment decisions in routine diabetes care.
Study context
- Population
- Individuals with T2D initiating GLP-1 RA
- Participants
- 69,194 individuals
- Exposure
- Early intensification (≤12 months) vs not intensified in first year (Model 1) or never intensified (>5 years; Model 2)
- Outcome
- Predictors, at GLP-1 RA prescription, of early intensification (≤ 12 months from GLP-1 RA initiation)
- Follow-up
- ≤ 12 months
Methodological detailsexplainable machine learning (logic learning machine), real-world
- Explainable machine learning (logic learning machine)
- Real-world
Main finding
As reported in the abstract
Discriminative ability for early intensification
| Group | AUC |
|---|---|
| Model 1 | 0.65 |
| Model 2 | 0.78 |
AUC
Model 10.65
Model 20.78
Observational data: associations, not effects.
Original publication
Identifying Predictors of Early Treatment Intensification in Individuals With Type 2 Diabetes Treated With GLP ‐1 Receptor Agonists: A Machine Learning–Based Analysis of a Large Real World Cohort. Diabetes Obes Metab. Published 9 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →