Physiological characterizations of patients with type 1 diabetes

Currently, there are no set physiological classifications of patients with type 1 diabetes (T1DM). Physiological groups of patients with T1DM developed using cluster analysis can inform diabetes management technologies and training methods. Using the FDA-approved simulator, we performed simulation runs on in silico adults (n=100) with T1DM and clustered the results based on sensitivity level to meal carbohydrate misestimation, basal insulin resistance, and glucose measurement sensor error. The clustering analysis performed revealed underlying groups of patients with T1DM. Based on the results, we developed an insulin therapy intervention to improve blood glucose outcomes. Glycemic control metrics combined with graphical analysis support the application of the intervention on the cluster with the highest sensitivities. In most cases, the intervention was effective in reducing the average daily risk index and the risk of hypoglycemia for these high-risk patients. In cases where the metrics did not show significant improvement, we found that using the intervention did no harm to the patient. With further research, we believe the clusters and interventions could greatly benefit future T1DM management.