A UMLS-based spell checker for natural language processing in vaccine safety

Table 4 Spell checker performance measures

PARAMETERS	TRAINING SET n = 12, 056			TEST SET n = 8,131
	Value	95% CI		Value	95% CI
		Lower	Higher		Lower	Higher
Sensitivity (%)	93	93	94	74	74	75
Specificity (%)	100	100	100	100	100	100
Recovery (%)	85	84	85	68	67	69
Positive Predictive Value	64	63	65	47	46	48
Regular expression transformations	1,217 (10%)			770 (9%)
Words corrected	105 (1%)			68 (1%)
Processing time per word (second)	0.07			0.06

ISSN: 1472-6947