Machine learning models for dementia screening to classify brain amyloid positivity on positron emission tomography using blood markers and demographic characteristics: a retrospective observational study

Table 2 Performance of the amyloid β positivity classification models using L2-regularized logistic regression

	ROC AUC	Sensitivity	Specificity	PPV	NPV	Accuracy
Model 0^a	0.67 (0.01)	0.64 (0.04)	0.62 (0.04)	0.52 (0.02)	0.74 (0.01)	0.63 (0.02)
Model 1^b	0.70 (0.01)	0.68 (0.04)	0.60 (0.04)	0.52 (0.02)	0.75 (0.02)	0.63 (0.01)
Model 2^c	0.70 (0.01)	0.60 (0.04)	0.68 (0.05)	0.56 (0.04)	0.74 (0.01)	0.66 (0.03)
Model 3^d	0.73 (0.01)	0.62 (0.04)	0.69 (0.04)	0.57 (0.03)	0.75 (0.01)	0.66 (0.02)
Model 4^e	0.76 (0.01)	0.64 (0.03)	0.75 (0.03)	0.62 (0.03)	0.77 (0.01)	0.70 (0.02)

Data are mean (standard deviation)
^aModel 0: demographic characteristics (age, sex, body mass index, years of education)
^bModel 1: Model 0 plus all MMSE subscores
^cModel 2: Model 0 plus blood test results (excluding ApoE4 phenotype) and the other demographic characteristics (medical history, current alcohol consumption, smoking status)
^dModel 3: Model 2 plus all MMSE subscores
^eModel 4: Model 3 plus ApoE4 phenotype
ApoE4 apolipoprotein E4, MMSE Mini Mental State Examination, NPV negative predictive value, PPV positive predictive value, ROC AUC receiver operating characteristic area under the curve

ISSN: 1758-9193