960-P: Personalized Diabetes Management Using Data from Continuous Glucose Monitors

Expansion of continuous glucose monitor (CGM) use in type 1 diabetes (T1D) patients has not fully translated into personalized, timely glucose management. We propose a standardized approach based on individual CGM data to set patient-specific disease-management targets and monitor progress. We developed tools to measure daily mean glucose (MG), time-in-range between 70-180 mg/dL (TIR), and glucose variability (GV); set patient-specific targets for these metrics; and detect deviation from these targets. We simulate the proposed approach on 2,668,063 hours of CGM data from 299 T1D patients.

To simulate a patient’s clinic visit after CGM use, their first 90 days of CGM data were used to generate personalized targets. To simulate patient disease management post-visit, the subsequent 14 days of data were used to generate alerts on each day which deviated from their target. The number of daily alerts for static targets (>65% TIR) vs. personalized targets (at least 5% improvement over the preceding 90-day average TIR) were compared.

The evaluation included all 248 patients with at least 104 days of data. Patient-specific personalized glucose targets varied significantly. Of these 248 patients, 150 had a historical mean TIR higher than the static target of 65%. For the 98 patients with TIR <65%, the number of alerts for deviations from target TIR were fewer for personalized than for static targets (7.14 vs. 8.42 during the 14-day window.

Personalized glucose targets based on each patient’s CGM history offer several advantages over static targets set without knowledge of the patient history. Personalized targets may be more achievable; for patients with high TIR targets to maintain performance may be more appropriate than lower static targets; and personalized targets may be set to produce fewer unnecessary alarms. Automated analyses of CGM data may empower care teams to support personalized, timely disease management.