BioMedical Engineering OnLine

Table 3 Confusion matrix for each segment window length of the test data from ALLSTAR database

From: Optimal length of R–R interval segment window for Lorenz plot detection of paroxysmal atrial fibrillation by machine learning

Segment window length, beats	Classification	Ground truth
		Training data			Test data
		Non-AF^a	AF^b	Accuracy	Non-AF^c	AFc	Accuracy
10	Non-AF	588,036	3341	0.99493	903,219	930	0.970
10	AF	2566	570,673		31,584	158,073
20	Non-AF	294,986	329	0.99892	458,259	692	0.983
20	AF	298	286,667		8750	79,173
50	Non-AF	118,089	11	0.99991	184,063	671	0.986
50	AF	10	114,770		2369	31,620
100	Non-AF	59,032	2	0.99,996	92,160	527	0.988
100	AF	3	57,371		807	15,838
200	Non-AF	29,505	0	0.99998	45,759	423	0.983
200	AF	1	28,670		509	7944
500	Non-AF	11,783	0	1.00000	18,006	281	0.973
500	AF	0	11,453		308	3230

AF atrial fibrillation, LP Lorenz plot
^aLP images obtained from pure (> 99%) SR cases
^bLP images obtained from pure (> 99%) CAF cases
^cAnnotations of LP images with the strict criteria

Back to article page

ISSN: 1475-925X

Contact us

Submission enquiries: journalsubmissions@springernature.com