Intermediary variables and algorithm parameters for an electronic algorithm for intravenous insulin infusion

Area of Science:

• Biomedical Engineering
• Computational Physiology
• Endocrinology

Background:

• Intravenous insulin infusion algorithms often use a two-step process involving maintenance rate (MR) estimation and next infusion rate (IR(next)) calculation.
• Current methods may not accurately estimate MR before achieving euglycemia, impacting the precision of insulin delivery.
• Blood glucose (BG) dynamics and patient-specific factors are crucial for effective insulin therapy management.

Purpose of the Study:

• To develop and validate a novel algorithm for intravenous insulin infusion that estimates the maintenance rate (MR) prior to achieving euglycemia.
• To compare the performance of the new algorithm's estimated infusion rate (IR(next)) against historically assigned rates.
• To propose practical recommendations for computerizing the advanced insulin infusion algorithm.

Main Methods:

• Computed a "maintenance rate cross step next estimate" (MR(csne)) using historical hyperglycemic data and previous insulin infusion rates (IR(previous)) and BG changes.
• Compared MR(csne) with the mean IR on historical stable intervals (MR(true)), an estimate of the biologic MR.
• Mathematically developed an expanded theory of the algorithm and compared hypothetically calculated MR(csne)-dependent IR(next) with historically assigned IR(next).

Main Results:

• MR(csne) and MR(true) were strongly correlated (R² = 0.88), despite median differences (2.7 vs. 3.2 units/h).
• Historically assigned and MR(csne)-dependent IR(next) during hyperglycemia were also correlated (R² = 0.87), with median differences (4.0 vs. 4.6 units/h).
• The algorithm establishes a fundamental equation relating IR, MR, and BG rate of change variables, incorporating unique natural parameters.

Conclusions:

• The described algorithm effectively estimates MR before euglycemia and computes MR-dependent IR(next) values.
• Key design features address glycemic variability and enhance safety by mitigating hypoglycemia risk.
• The method allows for specifying patient-condition-specific glycemic targets, promoting tailored insulin therapy.