Multimodal AI correlates of glucose spikes in people with normal glucose regulation, pre-diabetes and type 2 diabetes

Le résumé fourni par la source

Type 2 diabetes (T2D) is a multifaceted disease associated with several factors, including diet, genetics, exercise, sleep and gut microbiome. Current diagnostic and monitoring methods based on episodic assays like glycated hemoglobin (HbA1c) fail to capture its full complexity. Here, in a prospective cohort of 1,137 participants in the United States, we analyzed multimodal data from 347 deeply phenotyped individuals (174 normoglycemic, 79 prediabetic and 94 T2D). We found significant differences in the distribution of glucose spike metrics among different diabetes states, with longer expected time for spike resolution and higher values of nocturnal hypoglycemia in T2D. We identified significant correlations between mean glucose level and gut microbiome diversity, and between expected time for spike resolution and resting heart rate. Our multimodal glycemic risk profiles, validated in 1,955 normoglycemic and 114 prediabetic individuals from an independent cohort, improved risk stratification by highlighting substantial variability among individuals with the same value of HbA1c. Such a multimodal approach provides a detailed phenotype that can potentially improve T2D prevention, diagnosis and treatment, and is more informative than HbA1c. Multimodal data from 347 deeply phenotyped individuals including healthy, prediabetic individuals and individuals with T2D report remotely acquired patterns of glucose control via continuous glucose monitoring, and correlates them with diet and microbiome features and physiological signals, showing that these are able to discriminate individuals with T2D from control also in a large independent cohort.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.