Predicting the onset of diabetes-related complications after a diabetes diagnosis with machine learning algorithms
Authors' conclusion
Our ML models allow for the identification of individuals newly diagnosed with diabetes who are at increased risk of developing diabetes-related complications. The prediction performance varied across complications but within acceptable ranges as prediction tools.
Study context
- Population
- Individuals with a diagnosis of diabetes in Catalonia, administrative registers 2013–2017
- Participants
- 610,019 individuals
- Comparison
- Four machine learning (ML) algorithms: Logistic regression (LR), Decision tree (DT), Random Forest (RF) and XGB
- Outcome
- Prediction of diagnosis of diabetes-related complications (hypertension, renal failure, MI, CVD, retinopathy, congestive HF, cerebrovascular disease, peripheral vascular disease, stroke)
- Follow-up
- One-, two- and three-year post-diabetes diagnosis
Methodological detailslongitudinal, population-based administrative registers
- Longitudinal
- Population-based administrative registers
Performance
As reported in the abstract
Prediction performance across complications · three years since diabetes diagnosis
AUC range
60% (retinopathy) to 69% (congestive heart failure)
Accuracy range
60% (retinopathy) to 75% (hypertension)
Further findings
Predictors (Shapley additive explanations)
- Age associated with elevated risk for all complications except retinopathy
- Gender, other comorbidities, co-payment levels and age were the most relevant predictors
Original publication
Predicting the onset of diabetes-related complications after a diabetes diagnosis with machine learning algorithms. Diabetes Res Clin Pract. Published 16 September 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →