Performance of various classification models built for modelling diabetes and hypertension
| Type of classification | N for case/control | Classification accuracy at the best random classifier (%) | Classification accuracy for the different models used (%) | |||
|---|---|---|---|---|---|---|
| LR | SVM | k-NN | MDR | |||
| (i) Diabetes in general population | 2853/7779 | 73.2 | 80.7. | 81.3±1.3 | 78.6±0.85 | 78.30 |
| (ii) Diabetes in hypertensive population | 1322/1382 | 51.1 | 70.9 | 87.4±1.1 | 75.6±2.7 | 72.1 |
| (iii) Two-stage aggregate of (i) + (ii) − diabetes | 2853/7779 | 73.2 | N/A | 84.9 | 88.2 | N/A |
| (iv) Hypertension in general population | 6759/3873 | 63.6 | 82.4 | 82.4±0.6 | 80.0±0.8 | 80.9 |
| (v) Hypertension in diabetic population | 2427/5994 | 71.2 | 80.1 | 80.8±1.3 | 76.0±1.4 | 67.3 |
| (vi) Two-stage aggregate of (iv) + (v) − hypertension | 1322/1382 | 51.1 | N/A | 95.3 | 90.3 | N/A |
| Kuwait-specific data sets | ||||||
| (i) Diabetes in general population | 1334/4179 | 75.8 | 79.4 | 79.4 | 77.6 | 75.9 |
| (ii) Hypertension in general population | 3451/2062 | 62.6 | 80 | 79.9 | 76.8 | 77.9 |
| Asian-specific data sets | ||||||
| (i) Diabetes in general population | 976/2061 | 67.9 | 84.3 | 84.3 | 81.4 | 83.6 |
| (ii) Hypertension in general population | 1933/1104 | 63.7 | 86.8 | 86.8 | 83.3 | 83.8 |
LR, logistic regression; SVM, support vector machine; k-NN, k-nearest neighbours; MDR, Multifactor Dimensionality Reduction.