Identification of Cognitive Distraction Using Physiological Features for Adaptive Driving Safety Supporting System

Table 4

Comparing the detection capability with various combinations of individual features using AdaBoost.

Selected features	Cognitive; load	Average; accuracy	𝐹; value
Gaze	Conversation	67.6	68.8
	Arithmetic	66.7	67.7
	Both	72.8	80.5
Head	Conversation	84.5	86.2
	Arithmetic	78.3	81.2
	Both	84.0	89.0
Pupil	Conversation	85.0	84.2
	Arithmetic	86.2	86.5
	Both	87.0	90.0
RRI	Conversation	72.3	71.4
	Arithmetic	77.3	75.3
	Both	78.3	84.3
Gaze + head	Conversation	86.1	87.4
	Arithmetic	81.6
	Both	85.8	89.9
Gaze + pupil	Conversation	86.0	86.3
	Arithmetic	84.5	84.6
	Both	86.8	90.2
Gaze + RRI	Conversation	76.1	77.6
	Arithmetic	79.1	78.5
	Both	79.0	84.3
Head + pupil	Conversation	89.7	90.8
	Arithmetic	87.1	88.7
	Both	89.7	92.6
Head + RRI	Conversation	87.8	89.6
	Arithmetic	87.6	88.4
	Both	89.6	92.9
Pupil + RRI	Conversation	89.7	90.0
	Arithmetic	88.1	87.6
	Both	90.2	92.8
Gaze + head + pupil	Conversation	89.4	90.6
	Arithmetic	89.8
	Both	90.2	93.0
Gaze + head + RRI	Conversation	88.3	90.0
	Arithmetic	89.3
	Both	90.3	93.3
Gaze + pupil + RRI	Conversation	89.9	90.1
	Arithmetic	89.0	89.4
	Both	91.1	93.5
Head + pupil + RRI	Conversation	90.6	92.0
	Arithmetic	89.8	91.2
	Both	92.5	94.7
Gaze + head + pupil + RRI	Conversation	91.0	92.3
	Arithmetic	92.0	93.1
	Both	92.9	95.1