Visual Computing for Industry, Biomedicine, and Art

Table 11 Comparison of GPT-4 and ChatGPT on the radiology report plain language translation task

From: Translating radiology reports into plain language using ChatGPT and GPT-4 with prompt learning: results, limitations, and potential

		Good	Missing	Inaccurate	Incorrect
ChatGPT	Original prompt	55.2%	19.2%	24.8%	0.8%
	Optimized prompt	77.2%	9.2%	13.6%	0%
GPT-4	Original prompt	73.6%	8.0%	18.4%	0%
	Optimized prompt	96.8%	1.6%	1.6%	0%

Back to article page