Skip to main content

Table 3 Changes in CR interpretation performance stratified by emergency physicians’ work experience

From: Effect of deep learning-based assistive technology use on chest radiograph interpretation by emergency department physicians: a prospective interventional simulation-based study

Variable

Inexperienced physician

Experienced physician

Sensitivity

(95% CI)

Without DLCR

62.65 (55.88, 69.41)

61.13 (53.65, 68.62)

With DLCR

69.40 (65.14, 73.66)

68.21 (63.46, 72.97)

P value

 < 0.001

 < 0.001

Specificity

(95% CI)

Without DLCR

85.47 (79.09, 91.84)

91.98 (86.06, 94.01)

With DLCR

90.31 (87.21, 93.41)

92.76 (89.56, 95.96)

P value

0.004

0.577

Accuracy

(95% CI)

Without DLCR

70.23 (65.05, 75.40)

71.39 (65.79, 76.98)

With DLCR

76.35 (73.17, 79.53)

76.37 (72.83, 79.91)

P value

 < 0.001

 < 0.001

AUROC

(95% CI)

Without DLCR

0.741 (0.701, 0.780)

0.766 (0.721, 0.811)

With DLCR

0.799 (0.761, 0.837)

0.805 (0.780, 0.829)

P value

 < 0.001

0.079

  1. CR, chest radiograph; DLCR, deep learning-based assistive technology for chest radiograph; CI, confidence interval; AUROC, area under the receiver operating characteristic curve
Back to article page

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us