Design and Evaluation of a Novel Clinical Decision Support Algorithm for Weekly Insulin Efsitora Alfa

Digital health interventions can improve diabetes management and optimize insulin dosing. We describe the design and verification of a novel clinical decision support (CDS) algorithm for titration and management of insulin efsitora alfa (efsitora), an insulin receptor agonist designed for once-weekly administration. The CDS algorithm uses feedback control paired with efsitora pharmacokinetics (PK) and pharmacodynamics.

The CDS algorithm incorporates population PK and patient-specific data on fasting blood glucose (FBG) trends, prior insulin doses, hypoglycemia, and demographics. It applies proportional-derivative control, adaptive gain, and insulin activity estimations to guide long-acting basal insulin titration and reduce hypoglycemia. The CDS was assessed in a 52-week simulation using integrated glucose-insulin physiological models that evaluated treatment with efsitora and degludec in insulin-naïve and basal-switch virtual participants with type 2 diabetes (N = 10 000 each). The CDS (FBG target: 80-120 mg/dL) and conventional (Riddle; FBG target: ≤120 mg/dL) algorithms guided efsitora and degludec treatments, respectively. Glycated hemoglobin (HbA1c), FBG, and hypoglycemia outcomes were summarized.

Glycated hemoglobin and FBG reductions from baseline to week 52 were observed with the efsitora CDS (HbA1c: -1.1% to -1.3%; FBG: -50 to -71 mg/dL) and degludec conventional (HbA1c: -0.7% to -0.8%; FBG: -31 to -49 mg/dL) algorithms. Clinical decision support algorithm-guided dosing led to lower rates (events/participant/30 days) of nocturnal (0.19 vs 0.33) and 24-hour (1.21 vs 2.25) level 1 hypoglycemia compared with the conventional dosing algorithm.

The investigational CDS system may achieve comparable FBG and HbA1c reductions, and lower hypoglycemia rates, compared with conventional algorithms for daily basal insulin titration.