Table 2 Itemized questions used to rate the treatment recommendations for each MRI report. Two experienced orthopedic surgeons used Likert scales (1 to 5) or binary schemes (yes or no). Additionally, raters were asked to provide (free-text) comments for each patient.

From: A pilot study on the efficacy of GPT-4 in providing orthopedic treatment recommendations from MRI reports

Question to evaluate

Possible answers

The overall quality of the treatment recommendations is

Poor [1]—Fair [2]—Good [3]—Very good [4]—Excellent [5]

Treatment recommendations are based on scientific and clinical evidence

Strongly disagree [1]–Disagree [2]—Neutral [3]—Agree [4]—Strongly Agree [5]

Treatment recommendations are clinically useful and relevant

Strongly disagree [1]—Disagree [2]—Neutral [3]—Agree [4]—Strongly Agree [5]

Treatment recommendations are up to date

Yes–no

Treatment recommendations are consistent

Yes–no

Back to article page