Table 4.
The performance of Kernel SVM, decision tree, random forest, KNN, and Naïve Bayes using all six ocular instrumentation features.
| Model | Predicted | Actual classes | Accuracy (%) | Sensitivity (%) | Specificity (%) | AUROC (%) | |
|---|---|---|---|---|---|---|---|
| Healthy | PM | ||||||
| Kernel SVM | Healthy | 204 | 8 | 91.47 | 80.00 | 93.58 | 86.79 |
| PM | 14 | 32 | |||||
| Decision Tree | Healthy | 186 | 8 | 84.50 | 80.00 | 85.32 | 82.66 |
| PM | 32 | 32 | |||||
| Random Forest | Healthy | 204 | 10 | 90.70 | 75.00 | 93.58 | 84.29 |
| PM | 14 | 30 | |||||
| KNN | Healthy | 190 | 10 | 85.27 | 75.00 | 87.16 | 81.08 |
| PM | 28 | 30 | |||||
| Naïve Bayes | Healthy | 196 | 9 | 87.98 | 77.50 | 89.91 | 83.70 |
| PM | 22 | 31 | |||||
AUROC area under receiver operating characteristic curve, SVM support vector machine, PM pathologic myopia.