Table 4.
Clinical diagnosis machine learningmodel performance
| Naive Bayes | OA | MCI | D | OA | MCI | MCI | D | OA | D |
| Accuracy | 80.6% | 92.0% | 87.6% | 99.1% | |||||
| CI | 76.2 – 85.0% | 89.0 – 95.0% | 73.9 – 91.3% | 97.9 – 100% | |||||
| G-mean | 90.0% | 79.3% | 87.2% | 88.9% | 88.9% | 85.4% | 85.4% | 98.1% | 98.1% |
| Sensitivity | 85.3% | 60.7% | 91.3% | 80.8% | 97.9% | 78.4% | 93.2% | 96.2% | 100.0% |
| Specificity | 96.5% | 90.1% | 79.8% | 97.9% | 80.8% | 93.2% | 78.4% | 100.0% | 96.2% |
| Decision Tree | OA | MCI | D | OA | MCI | MCI | D | OA | D |
| Accuracy | 78.7% | 90.6% | 84.1% | 97.2% | |||||
| CI | 74.1 – 83.26% | 87.1 – 94.1% | 76.6 – 85.37% | 94.1 – 98.33 | |||||
| G-mean | 88.6% | 72.6% | 82.3% | 86.2% | 86.2% | 81.5% | 81.5% | 95.5% | 95.5% |
| Sensitivity | 80.8% | 58.8% | 90.1% | 75.0% | 99.0% | 73.2% | 90.7% | 92.3% | 98.8% |
| Specificity | 97.3% | 89.7% | 75.2% | 99.0% | 75.0% | 90.7% | 73.2% | 98.8% | 92.3% |
| Logistic Regression | OA | MCI | D | OA | MCI | MCI | D | OA | D |
| Accuracy | 87.6% | 84.6% | 88.3% | 96.4% | |||||
| CI | 83.9 – 90.7% | 80.6 – 88.6% | 84.7 – 91.9% | 94.3 – 98.5% | |||||
| G-mean | 90.7% | 82.9% | 95.4% | 81.8% | 81.8% | 83.9% | 83.9% | 93.5% | 93.5% |
| Sensitivity | 93.5% | 80.2% | 93.5% | 91.0% | 73.6% | 93.4% | 75.7% | 98.1% | 89.2% |
| Specificity | 88.0% | 92.1% | 97.3% | 73.6% | 91.0% | 75.7% | 93.4% | 89.2% | 98.1% |
Note: OA = older adult; MCI = mild cognitive impairment; D = dementia; CI = confidence interval; G-mean = geometric mean.