Detecting cardiovascular diseases using unsupervised machine learning clustering based on electronic medical records

Table 4 Distribution of CVD and non-CVD cases in each cluster with different predetermined number of clusters in the testing set

Predetermined No. clusters	Presence of CVD	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Cluster 5	Cluster 6	Cluster 7	Cluster 8	Total
k = 2	Non-CVD	907	8191							9098
	CVD	5213	1279							6492
	Total	6120	9470							15,590
	% of CVD	0.8518	0.1351
k = 4	Non-CVD	377	8105	616	0					9098
	CVD	306	1169	5017	0					6492
	Total	683	9274	5633	0					15,590
	% of CVD	0.4480	0.1261	0.8906	0
k = 8	Non-CVD	5978	476	518	0	0	311	1815	0	9098
	CVD	603	806	4381	3	1	213	485	0	6492
	Total	6581	1282	4899	3	1	524	2300	0	15,590
	% of CVD	0.0916	0.6287	0.8943	1.0000	1.0000	0.4065	0.2109

ISSN: 1471-2288