Detection of Periodic Leg Movements by Machine Learning Methods Using Polysomnographic Parameters Other Than Leg Electromyography

Table 3

Tabular results for 10-fold cross-validation for all folds and all model types (confusion matrix shows the classification of the cases in the test dataset). In confusion matrix, the columns denote the actual cases and the rows denote the predicted.

Fold number	Multilayer perceptron				-nearest neighbor				Random forest				Logistic regression
	Confusion matrix		Accuracy	RMSE	Confusion matrix		Accuracy	RMSE	Confusion matrix		Accuracy	RMSE	Confusion matrix		Accuracy	RMSE
1	378	55	0.8428	0.3562	432	38	0.9193	0.2841	445	40	0.9311	0.2584	406	67	0.8601	0.3352
	91	405			37	422			24	420			63	393
2	367	56	0.8299	0.3676	433	41	0.9182	0.286	449	60	0.9139	0.2655	404	78	0.8461	0.3458
	102	404			35	420			20	400			65	382
3	411	104	0.8256	0.3859	432	34	0.9236	0.2764	440	52	0.9128	0.2692	386	62	0.8439	0.3511
	58	356			37	426			29	408			83	398
4	406	96	0.8288	0.3718	430	41	0.9139	0.2934	434	43	0.916	0.2665	385	67	0.8375	0.3561
	63	364			39	419			35	417			84	393
5	378	56	0.8428	0.3619	426	32	0.9193	0.2841	428	53	0.8999	0.2821	410	87	0.8439	0.3512
	90	405			43	428			40	408			58	374
6	386	79	0.8267	0.3768	422	36	0.9117	0.2971	427	50	0.902	0.2881	391	70	0.8407	0.3573
	82	382			46	425			41	411			78	390
7	407	97	0.8297	0.3791	433	33	0.9267	0.2707	446	65	0.9063	0.2773	386	62	0.8448	0.3551
	61	363			35	427			22	395			82	398
8	375	63	0.8319	0.3746	429	33	0.9224	0.2785	437	50	0.9127	0.2728	379	64	0.8351	0.3602
	93	397			39	427			31	410			89	396
9	389	73	0.8362	0.3642	424	35	0.9149	0.2917	441	60	0.9063	0.2788	393	71	0.8427	0.3546
	79	387			44	425			27	400			75	389
10	397	83	0.8341	0.367	431	40	0.917	0.288	426	54	0.8966	0.2812	390	68	0.8427	0.348
	71	377			37	420			42	406			78	392
Mean			0.8329	0.3705			0.9187	0.285			0.9098	0.274			0.8437	0.3515
Standard deviation			0.0058	0.0084			0.0044	0.0077			0.0094	0.0086			0.0063	0.0068