Table 3.
Performance of machine learning models in classifying participants with stable cognition (SC) from participants with declining cognition (DC) in ADNI sample
| Model | Sensitivity, % (CI) | Specificity, % (CI) | PPV, % (CI) | NPV, % (CI) | Accuracy, % (CI) |
|---|---|---|---|---|---|
| Decision trees (DT) | 79.5 (72.1–85.6) | 21.5 (11.3–35.3) | 75.0 (71.8–78.0) | 26.2 (16.1–39.5) | 64.9 (57.8–71.4) |
| k-Nearest | 80.1 | 33.3 | 78.0 | 36.2 | 68.3 |
| Neighborhood (kNN) | (72.8–86.1) | (20.8–47.9) | (74.3–81.4) | (25.2–48.3) | (61.4–74.7) |
PPV, positive predictive value; NPV, negative predictive value.