Utility of ChatGPT in generating accurate client handouts for common veterinary internal medicine diseases.
Chat Generative Pre-trained Transformer's (ChatGPT) ability to generate client educational materials for use in practice is unknown.
To assess the educational quality of ChatGPT-generated client handouts for common internal medicine diseases. We hypothesized that ChatGPT can be used to efficiently generate easy-to-understand, accurate handouts for client education.
Small Animal Internal Medicine (SAIM) diplomates and pet owners were administered 2 separate electronic surveys.
Client handouts on diabetes mellitus (DM) and immune-mediated hemolytic anemia (IMHA) in dogs and inflammatory bowel disease (IBD) in cats were generated by using a standardized prompt in ChatGPT-3.5. Electronic surveys were distributed to both pet owners and American College of Veterinary Internal Medicine (ACVIM)-SAIM diplomates.
Pet owners (n = 50) reported a greater understanding of each disease process after reading the handouts for DM (Z = 5.865, P < .001), IMHA (Z = 5.953, P < .001), and IBD (Z = 5.508, P < .001). Median pet owner satisfaction scores (reported on a scale of 0 to 5, with 0 indicating poor satisfaction and 5 indicating maximal satisfaction) were 4 for DM, 4 for IMHA, and 5 for IBD. Many diplomates reported that they would use the handout on DM (n = 48/67; 71%), IBD (n = 47/62; 76%), and IMHA (n = 32/64; 50%) with either "minor" or "minimal to no" revisions.
Although some refinement is warranted, ChatGPT-3.5 was able to successfully generate client educational handouts for common internal medicine diseases.
To assess the educational quality of ChatGPT-generated client handouts for common internal medicine diseases. We hypothesized that ChatGPT can be used to efficiently generate easy-to-understand, accurate handouts for client education.
Small Animal Internal Medicine (SAIM) diplomates and pet owners were administered 2 separate electronic surveys.
Client handouts on diabetes mellitus (DM) and immune-mediated hemolytic anemia (IMHA) in dogs and inflammatory bowel disease (IBD) in cats were generated by using a standardized prompt in ChatGPT-3.5. Electronic surveys were distributed to both pet owners and American College of Veterinary Internal Medicine (ACVIM)-SAIM diplomates.
Pet owners (n = 50) reported a greater understanding of each disease process after reading the handouts for DM (Z = 5.865, P < .001), IMHA (Z = 5.953, P < .001), and IBD (Z = 5.508, P < .001). Median pet owner satisfaction scores (reported on a scale of 0 to 5, with 0 indicating poor satisfaction and 5 indicating maximal satisfaction) were 4 for DM, 4 for IMHA, and 5 for IBD. Many diplomates reported that they would use the handout on DM (n = 48/67; 71%), IBD (n = 47/62; 76%), and IMHA (n = 32/64; 50%) with either "minor" or "minimal to no" revisions.
Although some refinement is warranted, ChatGPT-3.5 was able to successfully generate client educational handouts for common internal medicine diseases.
Authors
Novak Novak, Reed-Jones Reed-Jones, Kedrowicz Kedrowicz, Birkenheuer Birkenheuer, Royal Royal, McCool McCool
View on Pubmed