Table B.2. Metrics about psychometric predicted validity using the most informative features of EPISODIX+S+P dataset, sorted by Cohen’s kappa coefficient.

EPISODIX+S+P DATA SET	Chi-square						ANOVA-F
			Precision-Recall area						Precision-Recall area
	F1	k↓	HC	MCI	AD	ALL	F1	k↓	HC	MCI	AD	ALL
GB: Gradient Boosting	0,92	0,89	0,85	0,81	0,95	0,87	0,96	0,95	0,94	0,86	0,98	0,93
ET: Extra Trees	0,88	0,83	0,76	0,75	0,92	0,81	0,93	0,90	0,89	0,79	0,95	0,88
RF: Random Forest	0,88	0,82	0,75	0,76	0,91	0,81	0,93	0,91	0,9	0,8	0,95	0,88
SVM: support Vector Machine	0,71	0,60	0,56	0,46	0,81	0,61	0,74	0,65	0,56	0,49	0,87	0,64
AB: Ada Boost	0,68	0,58	0,56	0,38	0,8	0,58	0,69	0,60	0,54	0,42	0,81	0,59
LR: Logistic Regression	0,61	0,54	0,5	0,29	0,84	0,55	0,59	0,50	0,58	0,21	0,83	0,54

Notes.

Episodix, semantic memory and procedural memory dataset (458 triples). Features: 10 most informative features, age and genre, using chi-squared and ANOVA F-value methods. ML algorithms: ET, Extra Trees classifier; GB, Gradient Boosting classifier; AB, Ada Boost classifier; SVM, Support Vector Machines; LR, Logistic regression; and RF, Random forest. Metrics: F1 score; Cohen’s Kappa (i.e., used as index to order best classification); average and precision-recall area distributed by cognitive group. Cross validation (cv-fold = 55).