Risk Analysis of Blood Glucose Data: A Quantitative Approach to Optimizing the Control of Insulin Dependent Diabetes

By Boris P. Kovatchev, Martin Straume, D. Cox, L. Farhy. 2000

133 citations recorded by the source corpus.

Patients with Insulin-Dependent Diabetes are continuously involved in a clinical optimization process: to maintain strict glycemic control without increasing their risk for hypoglycemia. This study offers quantitative tools for on-line assessment of the quality of this optimization, based on self-monitoring of blood glucose (SMBG). Ninety-six adults with Insulin Dependent Diabetes Mellitus (IDDM), age 35 ± 8 yrs., duration of diabetes 16 ± 10 yrs., HbAlc 8.6 ± 1.8%, 43 of whom had a recent history of severe hypoglycemia (SH), while 53 did not, used Lifescan One Touch II meters for 135 ± 53 SMBG readings over a month. For the following six months the subjects recorded occurrence of SH. The two patient groups, with and without a history of SH, did not differ in age, duration of diabetes, HbAlc, insulin units/day, average BG or BG variability. We suggest a computational procedure based on a symmetrization of the BG measurement scale and on a superimposed BG risk function, that allows for computation of two glycemic control markers: the Low BG Index (LBGI) and the High BG Index (HBGI). The LBGI is associated with SH: the LBGI and the rate of change of the BG risk, classified correctly 77% of the subjects with vs. without a history of SH and accounted for 46% of the variance of future SH. The HBGI, in combination with age, duration of diabetes and daily insulin dose, accounted for 57% of the variance of patients' glycosylated hemoglobin. We conclude that the LBGI and the HBGI are accurate on-line SMBG measures for patients' glycemic control.

Subfields: Endocrinology, Diabetes and Metabolism, Genetics, Medical & Health AI

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Endocrinology, Diabetes and Metabolism

1. Boris P. Kovatchev
2. Claudio Cobelli
3. Chiara Dalla Man
4. Philip S. Thomas

In Genetics

1. Claudio Cobelli
2. Boris P. Kovatchev
3. Jacob Steinhardt
4. Philip S. Thomas

In Medical & Health AI

1. Philip S. Thomas

Citations in the corpus

• The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models (2022, background): “In this case, the proxy reward is based off of a glycemic risk measure, which reflects the likelihood that a patient will suffer an acute hypoglycemic episode, developed by the medical community (Kovatchev et al., 2000).”
• A Dual Mode Adaptive Basal-Bolus Advisor Based on Reinforcement Learning (2019, unknown): “In addition, the low blood glycaemic index (i.e., risk of hypoglycaemia; LBGI), high blood glycaemic index (i.e., risk of hyperglycaemia; HBGI), the mean amplitude of glycaemic excursion (MAGE), and the total daily insulin intake (TDI) in units of insulin were estimated.”
• The UVA/PADOVA Type 1 Diabetes Simulator (2014, background): “where G(t) is plasma glucose concentration, I(t) is plasma insulin concentration, GIR(t) is exogenous glucose infusion rate, SG, SI, and p 2 are model parameters, and risk is defined as in (11-12).”