BMC Medical Informatics and Decision Making

Table 2 CPT prediction accuracy measures under different filter combinations and accuracy calculation approaches for each specialty

From: A tree based approach for multi-class classification of surgical procedures using structured and unstructured data

Algorithm:	Random Forest
Compare to:	CPT_RVUmax								CPT_set
Compare with predicted:	1 CPT				2 CPTs				1 CPT				2 CPTs
Filters:	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2
Cardio	45	53	57	62	49	59	58	71	50	54	60	67	54	62	60	72
General	68	78	77	87	71	73	76	81	69	78	79	88	71	73	76	81
Urology	52	62	56	71	55	67	63	72	54	63	56	71	57	69	65	74
OBGYN	57	74	64	81	59	74	65	82	58	74	67	82	61	74	68	84
Other	36	51	54	68	37	53	56	69	38	52	56	69	40	54	58	70
Average	52	64	62	74	54	65	64	75	54	64	64	75	57	66	65	76

Algorithm:	Complementary weight recalcultion
Compare to:	CPT_RVUmax								CPT_set
Compare with predicted:	1 CPT				2 CPTs				1 CPT				2 CPTs
Filters:	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2
	54	61	68	72	68	76	86	89	55	62	69	73	69	75	86	88
	70	79	79	88	74	81	84	90	71	79	80	88	76	83	84	91
	57	66	62	76	60	74	62	78	58	67	71	78	63	74	74	79
	63	77	70	83	68	80	76	84	66	79	72	85	75	83	82	90
	39	52	55	65	40	53	55	66	44	57	58	71	44	58	59	71
	57	67	67	77	62	73	73	81	59	69	70	79	65	75	77	84

Algorithm:	Neural Net Model
Compare to:	CPT_RVUmax								CPT_set
Compare with predicted:	1 CPT				2 CPTs				1 CPT				2 CPTs
Filters:	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2	C	F1	F2	F1F2
	39	45	48	54	39	46	49	54	39	46	50	56	39	47	51	57
	15	28	16	30	15	28	17	30	16	29	18	31	16	30	19	33
	20	24	22	26	20	24	22	26	22	24	23	28	23	26	23	28
	34	47	37	50	34	48	38	51	34	47	39	53	36	47	39	55
	17	20	22	25	17	20	23	26	17	21	23	27	17	22	25	27
	25	33	29	37	25	33	30	37	26	33	31	39	26	34	31	40

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com