A bimodal dataset for diabetes research

Abstract In recent years, with the continuous booming of diabetes patients, the research on diabetes and its complications, including pathogenesis, early diagnosis and therapeutic interventions, has attracted considerable attention. However, the lack of large-scale real datasets has significantly impeded its in-depth development. To address this challenge, we hereby disclose our diabetes dataset of 5,922 examples and 190 attributes, spanning across many detailed and well-curated clinical and demographic records, e.g., BMI, lifestyle factors, family history, glycemic control, insulin-related measures, lipid and metabolic profiles, which can shed light for reliable analyses of diabetes progression, complication risks and associated physiological and metabolic factors.