AI Tool Demonstrates High Accuracy in Genetic Counseling Questions

AI Tool Demonstrates High Accuracy in Genetic Counseling Questions - AI in Daily Life - News

Revolutionizing Genetic Counseling: A Groundbreaking Study on the Capabilities of ai in Gynecologic Oncology

At the Society of Gynecologic Oncology’s Annual Meeting on Women’s Cancer, a team of researchers from NYU Langone Health and its Perlmutter Cancer Center presented an intriguing study highlighting the remarkable potential of a generative ai tool in answering common genetic counseling questions. The researchers, led by Dr. Jharna M. Patel, showcased the capabilities of a specific ai model, ChatGPT, in providing accurate and comprehensive responses related to genetic syndromes counseling in gynecologic oncology.

Study Methodology and Findings

The researchers collaborated with gynecologic oncologists to identify 40 frequently asked questions by patients, aligning them with professional society guidelines. These queries were subsequently inputted into ChatGPT version 3.4, prompting the ai for answers. Following this, attending gynecologic oncologists assessed the accuracy and comprehensiveness of the ai-generated responses.

The results were noteworthy, with ChatGPT providing correct and comprehensive answers to 82.5% of the questions. The remaining responses either offered correct but incomplete information or were partially incorrect. Importantly, none of the answers were deemed completely incorrect.

High Accuracy in Genetic Counseling

The highest accuracy rates were demonstrated by questions pertaining to genetic counseling. Intriguingly, all 20 questions in this category elicited 100% accurate responses from ChatGPT. Moreover, questions addressing specific genetic disorders, such as testing for variations in the BRCA1 or BRCA2 gene and Lynch syndrome, displayed impressive accuracy rates, further emphasizing ChatGPT’s proficiency in this domain.

Dr. Marina Stasenko, the senior study author, emphasized the potential for further enhancements to the ai tool’s performance through continuous training on additional data and refining question sets. Despite its impressive results, it is envisioned that this technology will act as an assistant to human providers rather than a standalone solution in clinical settings.

These findings underscore the significant role of generative ai, as exemplified by ChatGPT, in addressing common genetic counseling queries within gynecologic oncology. With its high accuracy rates and potential for improvement, this ai tool is poised to enhance patient education and alleviate anxiety, ultimately contributing to more informed decision-making in cancer care.

Authors of the study, which include Drs. Patel, Stasenko, Hermann, Growdon, and Aviki, hail from various departments within NYU Langone Health and NYU Grossman School of Medicine. This interdisciplinary collaboration represents a significant step forward in advancing healthcare through innovative ai solutions.