Abstract
Purpose :
To implement use of an artificial intellignece (AI) language model in enhancing the readability of online websites dedicated specifically for patient education on the subject of cataract surgery.
Methods :
The first 50 websites identified with patient education material dedicated to cataract surgery were identified. Each website was categorized based on sponsor of the website, divided into "Institution", "Private Practice", or "Medical Organization". The online content of each website was extracted and assessed for readability using five standardized reading formulas, video content, and multi-language availability. AI language model was used to simplify existing English and Spanish text, with the goal of increasing readability. Converted text was manually assessed by the authors for acurracy and retention of information. Readability of English text was measured using multiple reading level measurements, including Flesch Reading Ease Score. Spanish text readability was assessed using the Fernandez-Huerta readability scale, the Spanish equivalent of the Flesch Reading Ease Score.
Results :
Sponsorship of the 50 websites included the following: 32 Institutions, 7 Private Practices, and 11 Medical Organizations. Extraction and grading of website content demonstrated average reading grade level of the material to be 11.68± 1.59, well above the recommended reading level for patients. After AI conversion, the overall average reading grade level was 7.94 ± 0.82 (p < 0.01). The first 10 search results had a lower average reading grade level (10.40 ± 1.59) and better Flesch Reading Ease score (57.51 ± 9.24) compared to the subsequent 40 results (11.99 ± 1.43, p = 0.01, 47.64 ± 8.59, p <0.01). English text was successfully converted into simplified Spanish with an average reading ease score of 61.17 ± 5.39 (8-9th grade level). Average native Spanish text reading ease score was improved from 57.41 ± 4.98 to 71.78 ± 5.24 (p < 0.01).
Conclusions :
The use of an AI language model conversion improved the readability of online patient education materials regarding cataract surgery, in both English and Spanish, with proper retention of information and accuracy. Use of AI language models may have potential use in other areas of ophthalmology and medicine in improving readability for patients.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.