2024 | Eric Steimetz, MD, MPH; Jeremy Minkowitz, DO; Elmer C. Gabutan, MD; Joan Ngichabe, MB ChB; Hagar Attia, MB BCH; Mordechai Hershkop, BA; Fatih Ozay, MD; Matthew G. Hanna, MD; Raavi Gupta, MD
Artificial intelligence (AI) chatbots can simplify pathology reports for patients, making them easier to understand. A study evaluated the ability of two chatbots, Bard and GPT-4, to explain pathology reports in simple terms and identify key information. The study used 1134 pathology reports from a hospital in Brooklyn, New York, and found that both chatbots significantly reduced the reading grade level of the reports and improved readability scores. Chatbot 1 simplified 87.57% of reports correctly, while chatbot 2 simplified 97.44% correctly. However, both chatbots had some inaccuracies and hallucinations, with chatbot 1 having 32 instances and chatbot 2 having 3 instances. The findings suggest that AI chatbots have the potential to explain pathology reports to patients, but they should not be used without review by a healthcare professional. The study highlights the need for further research to improve the accuracy and reliability of AI chatbots in clinical settings.Artificial intelligence (AI) chatbots can simplify pathology reports for patients, making them easier to understand. A study evaluated the ability of two chatbots, Bard and GPT-4, to explain pathology reports in simple terms and identify key information. The study used 1134 pathology reports from a hospital in Brooklyn, New York, and found that both chatbots significantly reduced the reading grade level of the reports and improved readability scores. Chatbot 1 simplified 87.57% of reports correctly, while chatbot 2 simplified 97.44% correctly. However, both chatbots had some inaccuracies and hallucinations, with chatbot 1 having 32 instances and chatbot 2 having 3 instances. The findings suggest that AI chatbots have the potential to explain pathology reports to patients, but they should not be used without review by a healthcare professional. The study highlights the need for further research to improve the accuracy and reliability of AI chatbots in clinical settings.