The purpose of this study was to investigate the association between AI-based digital scoring of the iris heart region and heart disease, and to evaluate its potential as a non-invasive diagnostic aid. A retrospective case-control study was conducted ...
The purpose of this study was to investigate the association between AI-based digital scoring of the iris heart region and heart disease, and to evaluate its potential as a non-invasive diagnostic aid. A retrospective case-control study was conducted with 366 participants, consisting of 161 clinically diagnosed heart disease patients and 205 age- and sex-matched controls. The ASKeye IRIS-Pro AI system was utilized to extract quantitative markers from the iris, specifically focusing on the heart and vessel regions. To validate diagnostic performance, three machine learning models—Logistic Regression, Support Vector Machine (SVM), and Random Forest—were trained and tested.
The results demonstrated that digital scores for the iris heart region were significantly higher in the heart disease group compared to the control group (p < 0.001). Crucially, ANCOVA analysis confirmed that the iris heart markers remained significant independent diagnostic factors even after controlling for age and sex, whereas vessel markers lost statistical significance. Among the classification models, the SVM achieved the highest accuracy of 86.4% (AUC 0.887), while the Random Forest model demonstrated the highest discriminative power with an AUC of 0.890 (Accuracy 84.5%) on the independent test set. Furthermore, feature importance analysis revealed that the heart region markers were the most critical predictors, accounting for 55.1% of the model's predictive power. These findings suggest that AI-based iris analysis provides an objective and reproducible method for heart disease screening, offering a valuable tool for early detection and preventive healthcare.