Table 3. Diagnostic Stability and Model Predictions Among Patients Who Were Initially Diagnosed With Dementia Within 2 Years of Their Baseline Assessment.
| Diagnosis status | Patients, No. (%) | |||||
|---|---|---|---|---|---|---|
| BDSI | CAIDE | LR | SVM | RF | XGB | |
| Correctly classified | ||||||
| Consistently diagnosed, model predicted to develop dementia | 536 (37.3) | 243 (16.9) | 694 (48.3) | 689 (47.9) | 477 (33.2) | 666 (46.3) |
| Diagnosis reversed, model predicted to stay dementia-freea | 91 (70.0) | 97 (74.6) | 92 (70.8) | 93 (71.5) | 109 (83.8) | 98 (75.4) |
| Misclassified | ||||||
| Consistently diagnosed, model predicted to stay dementia-free | 902 (62.7) | 1195 (83.1) | 744 (51.7) | 749 (52.1) | 961 (66.8) | 772 (53.7) |
| Diagnosis reversed, model predicted to develop dementiaa | 39 (30.0) | 33 (25.4) | 38 (29.2) | 37 (28.5) | 21 (16.2) | 32 (24.6) |
Abbreviations: BDSI, Brief Dementia Screening Indicator; CAIDE, Cardiovascular Risk Factors, Aging and Incidence of Dementia; LR, logistic regression; RF, random forest; SVM, support vector machine; XGB, gradient-boosted trees.
a
Patients considered as having their diagnosis reversed were initially diagnosed with dementia within 2 years of their baseline visit whose diagnosis was subsequently reversed to mild cognitive impairment or cognitively unimpaired within 2 years of further follow-up suggesting probable misdiagnosis.