Diabetes is a disorder traditionally subdivided into two types. Type 2 diabetes (T2D) is a chronic condition that affects how our body metabolizes sugar or glucose, inducing either resistance to the effects of insulin, or lack of its production in a way sufficient to maintain normal glucose levels. No cure exists for such disorder affecting populations that include adults as well as children. Control of body weight, diet, and exercise can help T2D management, complementing (or as an alternative to) medications or insulin therapy.

T2D is of interest to this work. The classification of diabetes depends primarily on age at onset and the presence or absence of conditions such as obesity, metabolic syndrome, insulin deficiency, and others. Several mechanisms can lead to diabetes, and these can be modified by genetic, lifestyle, and environmental factors. Clearly, all such factors make T2D a very heterogeneous disease, one for which many types of data should be analyzed for achieving superior precision of diagnoses and therapies. The identification first of the informative features and patterns within these complex “Big Data” sets and then of the linkages to outcome data may yield valuable insights into risk factors, diabetes history, and comorbidities, in turn advancing both prevention and management of the disease from a Precision Medicine (PM) perspective (see, for instance, Capobianco, 2017; Fitipaldi et al., 2018; Xie et al., 2018; Prasad and Groop, 2019). https://www.frontiersin.org/articles/10.3389/fdata.2019.00030/full#B21

Preo, Nicolo and Capobianco, Enrico, Significant EHR Feature-Driven T2D Inference: Predictive Machine Learning and Networks Frontiers in Big Data, September 27, 2019 https://www.frontiersin.org/article/10.3389/fdata.2019.00030