Machine Learning Creates a Simple Endoscopic Classification System that Improves Dysplasia Detection in Barrett’s Oesophagus amongst Non-expert Endoscopists

Table 1

Comparison of sensitivity and specificity between cases before and after acetic acid for all 3 experts. ND-BE: no dysplasia; D-BE: dysplasia.

	Expert 1 (%)	Expert 2 (%)	Expert 3 (%)	Total (%)
Before acetic acid
Sensitivity, % (95% C.I.)	94 (71–99)	76 (50–93)	94 (71–99)	88 (76–96)
Specificity, % (95% C.I.)	87 (66–97)	87 (66–97)	83 (61–95)	86 (75–93)
Negative predictive value, %	99.6	98.5	99.6	86 (75–93)
After acetic acid
Sensitivity, % (95% C.I.)	100 (48–100)	80 (28–99)	100 (48–100)	93 (68–99)
Specificity, % (95% C.I.)	71 (29–96)	85 (42–99)	57 (18–90)	71 (47–89)
Negative predictive value, %	100	98.8	100	71 (47–89)