Diagnostic Modelling
The clinical-risk-factor model outperformed the fundus-only model, and the combined model performed best. The logistic regression approach made the model more interpretable than many deep-learning retinal models because it used automated n…
1 sources - 6 claims
The clinical-risk-factor model outperformed the fundus-only model, and the combined model performed best. The logistic regression approach made the model more interpretable than many deep-learning retinal models because it used automated named retinal measurements. Bootstrap validation with 1000 iterations supported the combined model's stability. The study built fundus-only, traditional-risk-factor and combined diagnostic models. The model was designed to identify current CAD in suspected-angina patients rather than predict future cardiovascular events.