Table 7.

The precision of drowsiness detection was obtained with the XGBoost algorithm for each subject in the test set. The shaded green cells with bold text show the highest precision obtained for each subject. At the bottom, the best mean precision for each ratio index is marked in the same way.

Subject	Index1	Index2	θ/α	β/α	(θ + α)/β	θ/β	(θ + α)/(α + β)	γ/δ	(γ + β)/(δ + α)
	This Work		Eoh et al. [36]		Jap et al. [37]			da Silveira et al. [10]
tr04-0653	0.5769	0.6569	0.6742	0.5030	0.5133	0.6000	0.5862	0.5795	0.5914
tr05-0028	0.3818	0.4296	0.5178	0.5385	0.4762	0.4214	0.5364	0.5000	0.5862
tr05-0332	0.7483	0.9333	0.7905	0.5547	0.5321	0.7846	0.6709	0.8571	0.8824
tr07-0127	0.5401	0.6512	0.4722	0.5294	0.5231	0.5000	0.3600	0.4309	0.4207
tr08-0157	0.5812	0.6087	0.5158	0.5574	0.5000	0.5048	0.5153	0.5122	0.4886
tr09-0328	0.6147	0.5078	0.5650	0.5137	0.5000	0.5041	0.5678	0.5620	0.5372
tr09-0453	0.5398	0.5176	0.5436	0.5301	0.5248	0.5306	0.5538	0.5049	0.4721
tr12-0255	0.5316	0.5351	0.6054	0.5342	0.4826	0.5393	0.5545	0.5886	0.5389
tr12-0441	0.8000	0.5789	0.5633	0.5093	0.5249	0.5478	0.5785	0.6496	0.5231
tr13-0170	0.6333	0.5466	0.5724	0.5355	0.6056	0.6296	0.5313	0.6944	0.3478
tr13-0508	0.4337	0.5091	0.6331	0.5674	0.5443	0.6154	0.6140	0.5478	0.5352
tr13-0653	0.7048	0.6000	0.5177	0.5314	0.5419	0.5213	0.5845	0.5392	0.5037
Average	0.5905	0.5896	0.5809	0.5337	0.5224	0.5582	0.5544	0.5805	0.5356