โก Quick Summary
This study presents the development and evaluation of a diabetes-focused Generative Pretrained Transformer (GPT), achieving an impressive 91.7% accuracy in responding to diabetes-related inquiries. The tool aims to support healthcare professionals by providing accurate, well-sourced information in diabetes care.
๐ Key Details
- ๐ Dataset: 420 diabetes-related questions
- ๐งฉ Information Sources: 65 sources on diabetes care strategies and technologies
- โ๏ธ Technology: Generative Pretrained Transformer (GPT)
- ๐ Performance: Overall accuracy of 91.7%, with 100% rationale inclusion
๐ Key Takeaways
- ๐ Diabetes Help GPT provides accurate responses to diabetes-related questions.
- ๐ก High accuracy was noted in general diabetes knowledge and nutrition queries.
- ๐ฉโ๐ฌ Insulin-related questions had a slightly lower accuracy of 82.3%.
- ๐ 100% rationale inclusion ensures that responses are well-supported.
- ๐ Citations were included in 93.3% of responses, enhancing credibility.
- โ๏ธ Ethical compliance was maintained throughout the study.
- ๐ Developed through a transparent process with expert involvement.
- ๐ Designed to complement healthcare professionals’ expertise, not replace it.
๐ Background
Diabetes management is a complex field that requires accurate information and effective communication between healthcare providers and patients. The integration of advanced technologies, such as Generative Pretrained Transformers, into diabetes care can enhance patient education and support healthcare professionals in delivering high-quality care.
๐๏ธ Study
The study involved a systematic approach to developing the Diabetes Help GPT, which included a comprehensive literature review, selection and preprocessing of relevant information sources, prototype development, and rigorous evaluation using a set of 420 diabetes-related questions. This structured methodology ensured that the tool was both effective and reliable.
๐ Results
The Diabetes Help GPT demonstrated a remarkable overall accuracy of 91.7%. The tool excelled in providing rationale for its responses (100% inclusion), and it included citations in 93.3% of cases. Notably, the accuracy for insulin-related questions was slightly lower at 82.3%, indicating areas for potential improvement. The study also found significant variations in disclaimer and emoji usage based on question format.
๐ Impact and Implications
The successful development of the Diabetes Help GPT signifies a major step forward in utilizing AI for diabetes care. By providing accurate and well-sourced information, this tool can empower healthcare professionals and enhance patient education. The implications of such technology extend beyond diabetes, potentially influencing other areas of healthcare where accurate information dissemination is crucial.
๐ฎ Conclusion
The Diabetes Help GPT showcases the potential of AI in improving diabetes care through accurate and reliable information. As healthcare continues to evolve, tools like this can play a vital role in supporting professionals and enhancing patient outcomes. Continued research and development in this area are essential to fully realize the benefits of AI in healthcare.
๐ฌ Your comments
What are your thoughts on the integration of AI in diabetes care? We would love to hear your insights! ๐ฌ Leave your comments below or connect with us on social media:
Development and performance of a generative pretrained transformer for diabetes care.
Abstract
AIMS: To design and evaluate the performance of a diabetes-related Generative Pretrained Transformer (GPT).
METHODS: A prompt-engineered layer over GPT was developed in four stages: (1) literature review on GPT tools development; (2) selection and preprocessing of 65 information sources about diabetes care strategies, patient education, diabetes technologies, and cultural care, among others; (3) prototype development; and (4) final tool evaluation using 420 diabetes-related questions adapted from three validated instruments. Outcomes were accuracy, rationale, citations, disclaimers, and emoji exclusion. Statistical analyses included descriptive statistics, chi-square tests and bias assessment. Compliance with data protection regulations and ethical standards was ensured.
RESULTS: Diabetes Help GPT showed high overall accuracy (91.7โฏ%), with 100โฏ% rationale inclusion, 93.3โฏ% citations, 84.8โฏ% disclaimers, and minimal emoji use (13.3โฏ%). Accuracy was highest in general diabetes knowledge and nutrition questions; slightly lower in insulin-related items (82.3โฏ%). Disclaimer and emoji usage varied significantly by question format (pโฏ=โฏ0.026 and pโฏ<โฏ0.001). No accuracy bias was detected.
CONCLUSIONS: Diabetes Help GPT delivers accurate, well-sourced responses, supporting healthcare professionals in diabetes care. Unlike existing GPT models in medicine, it was developed through a transparent, expert-led process with curated content and iterative validation. It should complement, and not replace, professionals' criteria.
Author: [‘Garrido-Bueno M’, ‘Cruz-รlvarez PS’, ‘Pabรณn-Carrasco M’, ‘Romero-Castillo R’]
Journal: Diabetes Res Clin Pract
Citation: Garrido-Bueno M, et al. Development and performance of a generative pretrained transformer for diabetes care. Development and performance of a generative pretrained transformer for diabetes care. 2025; (unknown volume):112425. doi: 10.1016/j.diabres.2025.112425