Skip to main content

Table 3 Classification of dementia status with different algorithms and different combinations of neuropathology features. Prediction accuracy and standard error are listed for each feature when it is omitted from the multivariable classifier. Age and brain weight had the largest effect when dropped from the model, whilst the neuropathological parameters each had similar effects

From: Epidemiological pathology of Aβ deposition in the ageing brain in CFAS: addition of multiple Aβ-derived measures does not improve dementia assessment using logistic regression and machine learning approaches

Classifier TypeAge & Brain weightThal PhasePlaque scoreBraak StageCAANo features omitted
Logistic Regression0.6327 (+/−  0.0035)0.6774 (+/−  0.0033)0.6773 (+/−  0.0033)0.6776 (+/−  0.0033)0.6785 (+/−  0.0034)0.6773 (+/− 0.0033)
Decision Tree0.6327 (+/−  0.00400.6997 (+/−  0.00450.6929 (+/−  0.00450.7026 (+/−  0.00480.7011 (+/−  0.00470.7010 (+/−  0.0043)
LDA0.6344 (+/−  0.0032)0.6763 (+/−  0.0035)0.6709 (+/−  0.0034)0.6738 (+/−  0.0034)0.6760 (+/−  0.0.0036)0.6834 (+/−  0.0035)