It is estimated that by 2021, more than 500 million people worldwide will have diabetes.
A collaborative team of scientists from China and Singapore has developed an artificial intelligence (AI) tool for primary care of diabetes – the world's first multimodal large-scale language model (LLM) dedicated to managing this prevalent chronic disease.
As of 2021, it is estimated that more than 500 million people worldwide have diabetes, the majority of whom live in low- and middle-income countries, where there is often a shortage of trained primary care physicians and limited access to proper testing for diabetic retinopathy, a serious eye disease related to diabetes.
Researchers from Tsinghua University, Shanghai Jiao Tong University, and the National University of Singapore have developed a GPT-4-like system that can provide personalized diabetes management guidance for primary care physicians.
According to a study recently published in Nature Medicine, the image-plus-language platform, known as DeepDR-LLM, is designed to leverage the power of LLM and deep learning to provide a comprehensive solution for medical image diagnosis and delivering customized treatment recommendations.
The research team adapted the open-source LLM using 371,763 real-world management recommendations from 267,730 participants and then validated the imaging module using 21 datasets containing standard or portable retinal images from seven countries: China, Singapore, India, Thailand, the UK, Algeria, and Uzbekistan.
In a retrospective evaluation, the system performed as well as a family doctor in English and outperformed them in Chinese, according to the study.
Additionally, when it came to identifying diabetic retinopathy, primary care physicians' average accuracy increased from 81.0 percent without assistance to 92.3 percent with the system's support.
The researchers advocate for the adoption of this system in diabetic primary care protocols as it significantly improves the efficiency of both diagnosis and treatment, leading to improved health outcomes for diabetic patients.
