A recent study shows how artificial intelligence (AI) has significantly sped up the creation of software aimed at preventing diabetes.
Published on March 6 in the Journal of Medical Internet Research, the research focused on a specific AI technology, generative AI (GenAI), known for its ability to predict text sequences and generate realistic responses and summaries.
The study, conducted by NYU Langone Health researchers, applied ChatGPT to the development of a text-based program encouraging healthier eating and more exercise among patients. The results showed that using AI allowed a diverse team of medical and computer science experts to develop a diabetes prevention tool in just 40 hours, compared to over 200 hours without AI assistance.
Danissa Rodriguez, PhD, the study's corresponding author from NYU Langone's Department of Population Health and HiBRID Lab, emphasized, "ChatGPT improves communications between technical and non-technical team members to hasten the design of computational solutions to medical problems."
She suggested that if scalable, this approach could transform healthcare software design.
The study revealed that generative AI tools like ChatGPT could serve as effective translators between healthcare professionals and software engineers, facilitating the software development process. This approach, known as prompt engineering, requires skill in framing questions to the AI to achieve accurate responses.
In this case, clinical team members could input their ideas in simple English into ChatGPT, which then translated these ideas into technical specifications for the software engineers. While human developers were still necessary for the final coding, the AI significantly accelerated the overall design process.
Senior study author Devin Mann, MD, pointed to the potential of GenAI-assisted development to "democratize the design of healthcare software," enabling doctors and nurses to directly contribute to its creation without needing coding skills.
This study was a collaborative effort among several researchers at NYU Langone, and the project received funding from the National Institute of Diabetes and Digestive and Kidney Diseases.
