Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.
SUN-808 ADCES7 Guidelines and AI Chatbots: Do They Help Our Patients?
0
Zitationen
7
Autoren
2025
Jahr
Abstract
Abstract Disclosure: E. Pan: None. G. Wu: None. H. Lee: None. S. Sidhu: None. A. Sidhu: None. A. Ashok: None. A. Madala: None. Background: In 2020, 34.2 million US adults had diabetes and 88 million had pre-diabetes. The Association of Diabetes Care & Education Specialists (ADCES) lists seven categories of self-care behaviors for diabetics: Healthy Coping, Healthy Eating, Being Active, Taking Medication, Monitoring, Reducing Risk, and Problem Solving, collectively known as ADCES7. Purpose: Do responses by AI chatbots fulfill ADCES7 guidelines? Methods: Chatbots were prompted to generate important questions for each of the ADCES7 guidelines. The questions were then fed back to the chatbots and responses were recorded. The responses were then scored manually on a numerical scale from 1 to 5 based on accuracy and adherence to ADCES7 standards. Results: The results showed that average scores from the 5 LLMs regarding information about insulin was 1.77% lower than the average scores about diabetes monitoring. Chatbots generally performed well for questions about the seven main ADCES7 guidelines, but performed poorly about providing relevant information regarding insulin, insulin pumps, and GLP-1 analog medications. This highlights a critical gap in the current ADCES7 guidelines. Conclusion: The information in the ADCES7 handouts does not mention newer treatments or contain the word “insulin,” suggesting a gap in provided information and the need for an additional category focused on newer and more advanced treatment options, including insulin and GLP-1. Future research should also prioritize the development of more comprehensive guidelines and the enhancement of AI chatbot training to address knowledge gaps, ensuring that digital health tools can provide accurate, up-to-date information across all aspects of diabetes care, including the latest therapeutic advancements. The use of ADCES guidelines and information from other expert associations could be used as training texts for chatbots. Presentation: Sunday, July 13, 2025
Ähnliche Arbeiten
Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI
2019 · 8.336 Zit.
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead
2019 · 8.207 Zit.
High-performance medicine: the convergence of human and artificial intelligence
2018 · 7.607 Zit.
Proceedings of the 19th International Joint Conference on Artificial Intelligence
2005 · 5.776 Zit.
Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)
2018 · 5.476 Zit.