Evaluating the accuracy and reproducibility of ChatGPT-4 in answering patient questions related to small intestinal bacterial overgrowth

Table 1 Grading of responses generated by ChatGPT-4 to questions related to small intestinal bacterial overgrowth categorized by subgroup and overall

	%
Basic knowledge (n = 10)
Comprehensive	90
Correct but inadequate	10
Mixed with correct and incorrect data	0
Completely incorrect	0
Diagnosis (n = 5)
Comprehensive	60
Correct but inadequate	0
Mixed with correct and incorrect data	40
Completely incorrect	0
Treatment (n = 9)
Comprehensive	33.3
Correct but inadequate	0
Mixed with correct and incorrect data	33.3
Completely incorrect	33.3
Other (n = 3)
Comprehensive	33.3
Correct but inadequate	33.3
Mixed with correct and incorrect data	0
Completely incorrect	33.3
Overall (n = 27)
Comprehensive	59.3
Correct but inadequate	7.4
Mixed with correct and incorrect data	18.5
Completely incorrect	14.8

Table 2 Reproducibility of ChatGPT-4 responses overall and categorized by subgroup

	%
Overall (n = 27)	77.8
Basic knowledge (n = 10)	90
Diagnosis (n = 5)	80
Treatment (n = 9)	77.8
Other (n = 3)	33.3