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

Table 3 Distribution of CVD and non-CVD cases in each cluster with different predetermined number of clusters in the training 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	8402	73,478							81,880
	CVD	46,619	11,805							58,424
	Total	55,021	85,283							140,304
	% of CVD	0.8473	0.1384
k = 4	Non-CVD	3650	72,690	5538	2					81,880
	CVD	2889	10,743	44,792	0					58,424
	Total	6539	83,433	50,330	2					140,304
	% of CVD	0.4418	0.1288	0.8900	0
k = 8	Non-CVD	53,631	4367	4550	14	4	2	3025	16,287	81,880
	CVD	5554	7286	39,100	11	2	0	2255	4216	58,424
	Total	59,185	11,653	43,650	25	6	2	5280	20,503	140,304
	% of CVD	0.0938	0.6252	0.8958	0.4400	0.3333	0	0.4271	0.2056

ISSN: 1471-2288