150-LB: Software Modeling of Dorzagliatin for Artificial Intelligence Management of Type 2 Diabetes

• Published in: Diabetes (New York, N.Y.) Vol. 68; no. Supplement_1
• Main Authors: ALLEN, SYDNEY, BENNETT, BRITTANY, EILERMAN, BRADLEY, TESTA, LEONARD J.
• Format: Journal Article
• Language: English
• Published: New York American Diabetes Association 01.06.2019
• Subjects: Artificial intelligence; Beta cells; Clinical trials; Computer programs; Diabetes; Diabetes mellitus; Diabetes mellitus (non-insulin dependent); Drug dosages; Glucokinase; Hypoglycemia; Literature reviews; Population studies
• ISSN: 0012-1797; 1939-327X
• DOI: 10.2337/db19-150-LB

Introduction: The use of Artificial Intelligence to make clinical decisions is a strategy of increasing interest to improve the cost and effectiveness of diabetes management. GlucosePATH is software which uses massively parallel processing to sort through millions of possible combinations to medications specified to patient specific variables. As new medicine classes become available, the baseline characteristic of the medication must be defined. As dorzagliatin enters phase 3 trials, modeling the pharmacologic properties of the molecule will represent the first example of the glucokinase activator class. Methods: The first step in drug modeling defines the base characteristics of the molecule. Dorzagliatin is modeled mechanistically as a glucokinase activator which acts as a secretagog. This preventins mechanistic overlap and allows for modification of efficacy based on projected beta cell reserve. A dose of 75 mg daily was selected as it was the most efficacious dose in the phase 3 monotherapy trial. The next step in drug modeling is defining the clinical characteristics of the drug. At the dose defined, dorzagliatin is modeled with a HgA1c reduction of 1.2. This is modeled on the phase 3 monotherapy Weight change and hypoglycemia were modeled as neutral. The only adverse event of significance which was modeled was a mild penalty for fatty infiltration of the liver. All of these characteristics were drawn from the composites of phases 1 through 3 available on literature review. Results: Medications available in the market are described in terms of cost and insurance coverage. For the purpose of modeling dorzagilatin in a population, a cost of $300 monthly was entered as the low end of a new to market diabetes medication. Modeled on a population of 366 patients in the GlucosePATH database, the medicine was recommended in 4 patients. No significance difference between control and study population in mean HgA1c (both 7.3), cost ($629.99 vs. $629.79), or composite score out of 100 (82.65 vs. 82.64). Disclosure S. Allen: None. B. Bennett: None. B. Eilerman: Advisory Panel; Self; Dexcom, Inc., Medtronic. Speaker’s Bureau; Self; AbbVie Inc., AstraZeneca, Boehringer Ingelheim Pharmaceuticals, Inc., Corcept Therapeutics, Janssen Pharmaceuticals, Inc., Lilly Diabetes, Novo Nordisk Inc. Stock/Shareholder; Self; PATH Decision Support Software LLC. L.J. Testa: Stock/Shareholder; Self; PATH Decision Support Software LLC.