UP122 - Assessing the quality and readability of patient information available for Shockwave Lithotripsy (SWL): Human versus Generative Artificial Intelligence (GAI) comparison of resources and quality measurements – A benchmarking study from EAU Endourology

Published Date: 01st October 2026

Publication Authors: Harrison. NL

Introduction and Objectives
Shockwave Lithotripsy (SWL) is a non-invasive procedure designed to remove stones. Once a primary treatment for kidney stones, it is now mainly used for small, solitary, and uncomplicated urinary stones. As the number of patients with kidney stones rises, the online information available to patients and their caregivers must be high-quality and easy to understand. We aimed to evaluate the quality and readability of SWL patient information available online and compare it with leaflets created by Generative Artificial Intelligence (GAI).

Materials and Methods
The top 20 search results for four SWL-related search terms on Google, Yahoo, and Bing were collected, with duplicates excluded. GAI platforms, ChatGPT, Gemini, and DeepSeek, were used to produce a patient information leaflet. Two authors then assessed the quality using DISCERN scores and JAMA Benchmark Scores, and readability with the Flesch Reading Ease Score (FRES) and Flesch-Kincaid Grade Level (FKGL). ChatGPT was subsequently asked to evaluate the quality of the GAI-produced information using DISCERN.

Results
A total of 68 websites were included. The quality and readability assessments showed average websites with GAI-leaflet DISCERN scores of 52.4 (+/-8.91) and 58.7 (+/-5.03), JAMA Benchmark scores of 2.22 (+/-1.10) and 1 (+/-0), FRES scores of 52.1 (+/-13.8) and 51.5 (+/-8.05), and FKGL scores of 9.04 (+/-2.36) and 8.93 (+/-1.10), respectively. No significant differences were observed in DISCERN (p=0.224), FRES (p=0.652), and FKGL (p=0.866), but significant differences appeared in JAMA Benchmark scores (p=0.0298). Additionally, there was a significant difference in DISCERN scores between authors and ChatGPT (p<0.0001). Source Websites Websites AI-sources AI-sources p-value (websites vs AI) - Mean Standard Deviation Mean Standard Deviation -DISCERN Score (Range:15-75) For Quality Higher is better 52.4 8.91 58.7 5.03 0.224 DISCERN graded by AI 66.9 6.54 67 4.36 0.961 JAMA Benchmark (Range: 0-4) For authorship, attribution, disclosure, and currency Higher is better 2.22 1.10 1 0 0.0298 FRES (Range:0-100) Readability Higher is better 52.1 13.8 51.5 8.05 0.652 FKGL (Range: 0-12) Readability Lower is better 9.04 2.36 8.93 1.10 0.866

Conclusion
There is an apparent lack of high-quality SWL-related information that meets the average global readability standard. As the general public increasingly uses GAI resources and their training datasets are sourced from online websites, there is a need to enhance the quality of resources available to patients and carers. This not only helps them make informed decisions but also supports their psychological well-being during treatment, leading to greater comfort and improved outcomes.

Baskaran, R; Harrison, NL et al. (2026). UP122 - Assessing the quality and readability of patient information available for Shockwave Lithotripsy (SWL): Human versus Generative Artificial Intelligence (GAI) comparison of resources and quality measurements – A benchmarking study from EAU Endourology. European Urology Open Science. 92(S1), pp.S248-S249. [Online]. Available at: https://dx.doi.org/10.1016/S2666-1683%2826%2902022-7 [Accessed 15 October 2026]