. 2020 Nov 27;8(11):e19416. doi: 10.2196/19416

Table 4.

A comparison of the primary physician performance with the human-algorithm integration (HAI) performance, divided by physician experience (n=30); the Wilcoxon signed-rank test was used to compare the physician-alone performance and the HAI performance.

Primary physician characteristics and performance			Novice group (n=18)		Experienced group (n=12)		P value
Age in years, median (IQR)			27.00 (27.00-28.00)		32.00 (30.75-34.25)		<.001^a
Years of experience, median (IQR)			2.00 (2.00-3.00)		5.00 (4.00-6.25)		< .001^a
Performance evaluation
	Human-algorithm agreement, κ, median (IQR)
		Physician alone		0.66 (0.62-0.72)		0.69 (0.64-0.77)		.330
		HAI		0.77 (0.71-0.80)		0.82 (0.79-0.82)		.008^a
		Paired test, P value		.0001^a		.001^a
	Accuracy, median (IQR)
		Physician alone		0.90 (0.82-0.92)		0.90 (0.89-0.96)		.279
		HAI		0.94 (0.91-0.97)		0.97 (0.95-0.98)		.020 ^a
		Paired test, P value		.0023^a		.0032^a
	Sensitivity, median (IQR)
		Physician alone		0.91 (0.83-0.95)		0.98 (0.94-1.00)		.017^a
		HAI		0.97 (0.94-1.00)		1.00 (0.97-1.00)		.043^a
		Paired test P value		.0028^a		.0313^a
	Specificity, median (IQR)
		Physician alone		0.89 (0.84-0.94)		0.86 (0.81-0.94)		.733
		HAI		0.94 (0.88-0.96)		0.96 (0.92-0.98)		.215
		Paired test, P value		.1067		.0049^a

^aP value is statistically significant.