Table 3.

Performance of the trained supervised algorithms on the test set (20% of the whole dataset, i.e., 655 data points in case 1 (140 frames) and 96 data points in case 2 (20 first frames)). Performance was assessed by the number of true negatives (TN), true positives (TP), false negatives (FN), false positives (FP), accuracy, area under the curve (AUC), and 95% confidence interval (CI). The last column (#) orders the algorithms from best performance (1) to worse performance (6).

Case	Algorithm	TN	TP	FN	FP	TN (%)	TP (%)	FN (%)	FP (%)	Accuracy (%)	AUC	CI	#
140 frames	Logistic Regression	316	164	76	99	48	25	12	15	73.28	0.71	[0.70,0.77]	5
	K-nearest Neighbors (K=8)	348	198	44	65	53	30	7	10	83.34	0.82	[0.81,0.86]	1
	Kernel Support Vector Machine	340	166	52	97	52	25	8	15	77.25	0.75	[0.74,0.81]	4
	Naïve Bayes	326	126	66	137	50	19	10	21	69.00	0.67	[0.67,0.73]	6
	Decision Tree Classification	327	188	65	75	50	29	10	11	78.63	0.77	[0.76,0.82]	3
	Random Forest Classification	340	179	52	84	52	27	8	13	79.24	0.77	[0.76,0.82]	2
20 first frames	Logistic Regression	49	28	10	9	51	29	10	9	80.2	0.79	[0.72,0.88]	5
	K-nearest Neighbors (K=5)	53	36	6	1	55	38	6	1	92.7	0.94	[0.88,0.98]	2
	Kernel Support Vector Machine	49	27	10	10	51	28	10	10	79.2	0.78	[0.71,0.87]	6
	Naïve Bayes	54	28	5	9	56	29	5	9	85.4	0.84	[0.78,0.93]	4
	Decision Tree Classification	54	36	5	1	56	38	5	1	93.8	0.94	[0.89,0.99]	1
	Random Forest Classification	53	36	6	1	55	38	6	1	92.7	0.94	[0.88,0.98]	2