Machine Learning for the Preliminary Diagnosis of Dementia

Table 4

Overall performance of the diagnostic models.

Algorithm	Feature selection	Accuracy	Precision	Recall	F-measure
Random Forest	Relief	0.78	0.80	0.78	0.78
	Information Gain	0.78	0.79	0.78	0.78
	Random Forest	0.76	0.77	0.76	0.76
AdaBoost	Relief	0.77	0.78	0.77	0.77
	Information Gain	0.77	0.78	0.77	0.77
	Random Forest	0.76	0.76	0.76	0.76
LogitBoost	Relief	0.80	0.75	0.80	0.76
	Information Gain	0.78	0.73	0.78	0.74
	Random Forest	0.76	0.77	0.76	0.74
MLP	Relief	0.81	0.75	0.81	0.77
	Information Gain	0.79	0.73	0.79	0.75
	Random Forest	0.78	0.76	0.78	0.76
Naïve Bayes	Relief	0.79	0.74	0.79	0.75
	Information Gain	0.81	0.82	0.81	0.81
	Random Forest	0.77	0.80	0.77	0.78
SVM	Relief	0.80	0.74	0.80	0.76
	Information Gain	0.79	0.73	0.79	0.75
	Random Forest	0.76	0.74	0.76	0.75

Results were obtained after using the feature selection.