Table 6.

Accuracy with Random Forest

Input data table	Used text	Class	Per class accuracy	Accuracy
ScispaCy	Abstract	prevention	0.76
		treatment	0.69	0.74
		diagnosis	0.75
		case_report	0.74
TF-IDF	Abstract	prevention	0.96
		treatment	0.92	0.92
		diagnosis	0.90
		case_report	0.89
BOW	Abstract	prevention	0.93
		treatment	0.90	0.90
		diagnosis	0.92
		case_report	0.87
ScispaCy	Title	prevention	0.64
		treatment	0.50	0.57
		diagnosis	0.62
		case_report	0.47
TF-IDF	Title	prevention	0.83
		treatment	0.81	0.80
		diagnosis	0.81
		case_report	0.77
BOW	Title	prevention	0.71
		treatment	0.66	0.70
		diagnosis	0.78
		case_report	0.64
ScispaCy	Title_and_Abstract	prevention	0.81
		treatment	0.76	0.79
		diagnosis	0.80
		case_report	0.79
TF-IDF	Title_and_Abstract	prevention	0.96
		treatment	0.92	0.92
		diagnosis	0.91
		case_report	0.9
BOW	Title_and_Abstract	prevention	0.93
		treatment	0.90	0.91
		diagnosis	0.92
		case_report	0.87
BOW	Title_and_Abstract_and_	prevention	0.82
	Bibliometric Features	treatment	0.73	0.73
		diagnosis	0.67
		case_report	0.72
Input data table	Used text	Class	Per class accuracy	Accuracy
TF-IDF	Title_and_Abstract_and_	prevention	0.96
	Bibliometric Features	treatment	0.93	0.92
		diagnosis	0.90
		case_report	0.89