Predicting diabetic retinopathy based on biomarkers: Classification and regression tree models
Authors' conclusion
Based on the CART model analysis, routine serum biochemistry markers such as UCr, cortisol, and CRP, along with their respective thresholds, have been identified as potentially useful for identifying risk factors associated with DR at different the prevalence of DR rates. These findings might provided clinicians with a useful reference for the preliminary assessment of DR severity.
Study context
- Population
- Patients with diabetic mellitus (DM)
- Participants
- 781 patients
- Comparison
- DM patients with diabetic retinopathy (DR) vs without DR
- DRn = 395
- Without DRn = 386reference
- Outcome
- Prevalence of DR by CART-identified thresholds
Methodological detailsClassification and regression tree (CART) machine learning models, univariate and multivariate logistic regression
- CART machine learning models
- Univariate and multivariate logistic regression
Main finding
As reported in the abstract
Indicators identified as most critical, of 96 clinical test indicators
11
DR prevalence with UCr ≥ 127.75 mg/dl and cortisol ≥ 318.8 µg/l
100 %
DR prevalence with UCr < 127.75 mg/dl and cortisol ≥ 226.75 µg/l
79.6 %
Observational data: associations, not effects.
Further findings
Subgroup analysis with CART model
- Increased C-reactive protein (CRP), decreased 30-minute C-peptide (C-P1), and elevated cortisol, urinary protein (PRO) and UCr identified as risk factors for DR
Original publication
Composed by AI from the published abstract and checked against it. How Briefs are made →