Quick Search:

OC01.02: Accuracy and readability of patient‐focused information on obstetrics ultrasound imaging from online sources versus ChatGPT‐generated

Piersson, Albert ORCID logoORCID: https://orcid.org/0000-0001-9167-0269 and Dzefi‐Tettey, K. (2023) OC01.02: Accuracy and readability of patient‐focused information on obstetrics ultrasound imaging from online sources versus ChatGPT‐generated. Ultrasound in Obstetrics & Gynecology, 62 (S1). pp. 1-2.

Full text not available from this repository.

Abstract

Objectives

To investigate the accuracy and readability of patient-focused Information on obstetrics ultrasound imaging.

Methods

We blinded expert reviewers to evaluate the accuracy of ChatGPT-generated information compared with the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG), Radiologyinfo.org, American College of Obstetricians and Gynecologists (ACOG), British Medical Ultrasound Society (BMUS) and the UK National Health Service (NHS) answers on obstetrics ultrasound. We further evaluated the average Flesch-Kincaid readability grade level and Flesch reading ease score for each website and compared with the National Institutes of Health (NIH) and American Medical Association (AMA) recommended readability parameters.

Results

An overall 100% interrater agreement was achieved for the answers to the online sourced information and 95% for ChatGPT outputs. The average grade reading level of information on ISUOG, Radiologyinfo.org, ACOG, BMUS, NHS, and ChatGPT was 8.62 ± 3.6; 95% CI [6.46, 10.77], 12.21 ± 4.7; 95% CI [8.87, 15.6], 12.8 ± 2.9; 95% CI (9.2, 16.4], 10.5 ± 3.6; 95% CI [8.2, 12.8], and 10.5 ± 3.6; 95% CI [8.2, 12.8], respectively. On the other hand, the average reading ease score of information exceeded the AMA recommended parameters as follows: ISUOG, Radiologyinfo.org, ACOG, BMUS, NHS, and ChatGPT was 59.7 ± 14.6; 95% CI [50.9, 68.5] (“fairly difficult high school”), 38.67 ± 16.5; 95% CI [26.9, 50.5] (“difficult college”), 43.5 ± 13.2; 95% CI [26.9, 60.0] (“difficult college”), 46.9 ± 19.7; 95% CI [34.3, 59.4] (“difficult college”), and 65.6 ± 11.0; 95% CI [59.0, 72.3] (“difficult college”], respectively.

Conclusions

ChatGPT provides accurate patient-focused information on obstetrics ultrasound imaging. However, the readability of currently available patient-focused information on obstetrics ultrasound imaging may be incomprehensible to the general public.

Item Type: Article
Status: Published
DOI: 10.1002/uog.26322
School/Department: School of Science, Technology and Health
URI: https://ray.yorksj.ac.uk/id/eprint/12383

University Staff: Request a correction | RaY Editors: Update this record