Skip to main content

Table 4 Quality of machine learning suggestions: proportion of machine learning judgements (i.e. ‘low’ or ‘high/unclear’) overturned by the user

From: Machine learning to help researchers evaluate biases in clinical trials: a prospective, randomized user study

 

Overall

(n = 328)

RSG

(n = 82)

AC

(n = 82)

BPP

(n = 82)

BOA

(n = 82)

Judgements changed

27 (9%)

7 (9%)

7 (9%)

6 (7%)

7 (9%)