Fig. 1: Differences in surgical consent form readability and linguistic parameters before and after simplification.

Differences in surgical consent form readability before and after simplification mediated by GPT-4. Median and interquartile range for each variable are plotted. P-values reported correspond to results from nonparametric Mann-Whitney tests. For (A–C), a distinct color was used to label each individual institution (n = 15). A Differences in reading time. B Differences in Flesch-Kincaid Reading Level. C Differences in Flesch Reading Ease score. D Differences in other linguistic variables before (red) and after (blue) simplification. Due to differences in scale between variables, all results were reported as percentages (0–100%) of a predetermined constant: 5000 for total characters, 900 for total words, 40 for total sentences, 30 for total paragraphs, 6 for characters per word, 25 for words per sentence, 2.5 for sentences per paragraph, and 4000 for average word rarity. E Histogram visualizing changes in the distribution of word frequency ranking of consent form text before (blue) and after (red) simplification mediated by GPT-4. Word frequency in terms of rank within the English language is plotted on the x-axis, with higher rank denoting increased rarity and ranks 10,000 and above combined into a single category. Solid lines denote the distribution of word frequency ranking following fitting the data to a γ distribution.