Unveiling the sound of the cognitive status: Machine Learning-based speech analysis in the Alzheimer’s disease spectrum

Table 2 Discriminatory capacity of cognitive composites for differentiating clinical phenotypes using subjects with subjective cognitive decline (SCD) as reference

Composite	Accuracy		Sensitivity		Specificity		AUC
Composite	ADD	MCI	ADD	MCI	ADD	MCI	ADD	MCI
Attention	0.87	0.74	0.86	0.74	0.89	0.75	0.95	0.84
Executive function	0.97	0.84	0.97	0.83	0.98	0.91	0.99	0.93
Language	0.96	0.81	0.95	0.79	0.98	0.91	0.99	0.91
Memory	0.98	0.81	0.98	0.80	0.97	0.87	0.99	0.90
Visuospatial function	0.95	0.79	0.95	0.77	0.98	0.92	0.99	0.91

Mean value obtained over the test set from ten repetitions of tenfold cross-validation. A logistic regression model was employed to predict clinical phenotypes. The composite scores were considered as independent variables
Abbreviations: AUC, area under the curve; MCI, mild impairment; ADD, Alzheimer’s disease dementia

ISSN: 1758-9193