Diabetics, or others who monitor their sugar intake, may look at cookies and wonder, “How will eating this affect my blood sugar levels?” Generative AI models can now predict answers.
Tel Aviv-based startup Pheno.AI and researchers at NVIDIA's Weizmann Institute of Science led the development. Guruformeran AI model that can predict an individual's future blood sugar levels and other health indicators based on past blood sugar monitoring data.
Data from continuous blood glucose monitoring could help diagnose patients with prediabetes or diabetes more quickly, the researchers say. Harvard Health Publishing and New York University Langone Health. GluFormer's AI capabilities further enhance the value of this data, helping clinicians and patients spot anomalies, predict clinical trial results, and predict health status up to four years in advance.
Researchers also found that by adding dietary intake data to the model, GluFormer can also predict how a person's blood sugar levels will respond to specific foods or dietary changes, allowing for precision nutrition. I showed that.
Accurately predicting blood sugar levels in people at high risk of developing diabetes will enable physicians and patients to adopt preventive treatment strategies sooner, improving patient outcomes and reducing the economic impact of diabetes. There is a possibility that $2.5 trillion worldwide by 2030.
AI tools like GluFormer have the potential to help hundreds of millions of adults with diabetes. The disease currently affects approximately 10% of adults worldwide, but this number is potentially increasing Double by 2050 Affecting over 1.3 billion people. One of the. 10 main causes of death Side effects such as kidney damage, vision loss, and heart disease have been reported around the world.
GluFormer is a transformer model, a type of neural network architecture that tracks relationships in sequential data. This is the same architecture as OpenAI's GPT model. In this case, it will generate blood sugar values instead of text.
“Medical data, especially continuous blood glucose monitoring, can be thought of as a series of diagnostic tests that track biological processes throughout the lifespan,” said Gal Chechik, senior director of AI research at NVIDIA. “We found that a transformer architecture developed for long text sequences can take a series of medical tests and predict the results of the next test. Develop or learn something.”
The model was trained on 14 days of glucose monitoring data from more than 10,000 non-diabetic study participants, with data collected every 15 minutes through a wearable monitoring device. Data was collected as part of: Human Phenotype Projectan initiative by Pheno.AI, a startup aiming to improve human health through data collection and analysis.
“Two important factors came together to make this research possible: the maturation of generative AI technology powered by NVIDIA and the collection of large-scale health data by the Weizmann Institute,” the paper says. said NVIDIA researcher Guy Lutzker, lead author of . and Ph.D. student at the Weizmann Institute of Science. “This puts us in a unique position to extract interesting medical insights from the data.”
The research team validated GluFormer across 15 other datasets and found it generalizable to predict health outcomes in other groups, including prediabetes, type 1 and type 2 diabetes, gestational diabetes, and obesity. .
They used the following clusters: NVIDIA Tensor Core GPU Accelerate model training and inference.
Beyond blood sugar levels, GluFormer can predict medical values such as visceral adipose tissue, which is a measure of the amount of body fat around organs such as the liver and pancreas. Systolic blood pressure associated with diabetes risk. Apnea-hypopnea index is a measurement of sleep apnea associated with type 2 diabetes.
please read GluFormer research paper on Arxiv.
