993-P: Comparison of Artificial-Intelligence Decision Support for Multiple Daily Injection Therapy with Automated Insulin Delivery after 3 Months of Use in Silico
| العنوان: | 993-P: Comparison of Artificial-Intelligence Decision Support for Multiple Daily Injection Therapy with Automated Insulin Delivery after 3 Months of Use in Silico |
|---|---|
| المؤلفون: | Leah M. Wilson, Robert Dodier, Peter G. Jacobs, Clara Mosquera-Lopez, Joseph El Youssef, Nichole S. Tyler, Wade Hilts, Jessica R. Castle |
| المصدر: | Diabetes. 69 |
| بيانات النشر: | American Diabetes Association, 2020. |
| سنة النشر: | 2020 |
| مصطلحات موضوعية: | Decision support system, Multiple dose regimen, business.industry, Endocrinology, Diabetes and Metabolism, In silico, Internal Medicine, Insulin delivery, Medicine, business, Bioinformatics |
| الوصف: | While automated insulin delivery (AID) systems are now commercially available, over 40 percent of people with type 1 diabetes manage their insulin with multiple daily injection therapy (MDI). AID systems can improve time-in-range in adults, however it is not clear if optimal MDI therapy can achieve the same glycemic targets. We have previously shown that a K-nearest-neighbors decision support system (KNN-DSS) for MDI can deliver safe recommendations and achieves high agreement with endocrinologists. Here, we report glycemic outcomes as we compare KNN-DSS with two artificial pancreas systems: fading-memory-proportional-derivative (FMPD), and model-predictive control (MPC). We utilized a platform accepted by U.S. Food and Drug Administration for the evaluation of artificial pancreas technologies, the UVA-Padova simulator. In silico, 100 adults with varying insulin sensitivities, weight, and insulin dosing behaviors underwent a 3-arm study comparing KNN-DSS, FMPD, and MPC algorithms. Each UVA-Padova subject received a unique study meal pattern that was repeated across study arms. The two primary outcome measures were % time-in-range and % time-in-hypoglycemia. Secondary outcome measures included LGBI and HGBI risk index. The KNN-DSS was able to achieve 79.6 ± 15.4 %time-in-range after 12 weeks compared with 83.6 ± 8.3% for the MPC and 84.2 ± 9.24% for the FMPD. The KNN-DSS achieved comparable time below 70 mg/dL (0.79%, [0 1.6]) with the MPC-AP (0.62%, [0 1.8]) and FMPD (0%, [0 0.72]). The KNN-DSS also achieved comparable LGBI (2.55) with the MPC-AP (2.68). This simulation assumes perfect adherence to insulin dosing regimens, and while % time-in-range may be higher than expected, the results indicate that decision support in MDI may yield comparable glycemic outcomes with automated insulin delivery systems if people are compliant. Results of this study will be used to guide the design of an upcoming human clinical trial. Disclosure N.S. Tyler: None. W. Hilts: None. C.M. Mosquera-Lopez: None. L.M. Wilson: None. R. Dodier: None. J. El Youssef: None. J.R. Castle: Advisory Panel; Self; Novo Nordisk Inc. Consultant; Self; ADOCIA, Dexcom, Inc., Zealand Pharma A/S. Other Relationship; Self; Pacific Diabetes Technologies, Roche Diabetes Care. P.G. Jacobs: Consultant; Self; SFC Fluidics. Research Support; Self; Dexcom, Inc. Stock/Shareholder; Self; Pacific Diabetes Technologies. Other Relationship; Self; AgaMatrix. Funding The Leona M. and Harry B. Helmsley Charitable Trust (2018PG-T1D001) |
| تدمد: | 1939-327X 0012-1797 |
| الوصول الحر: | https://explore.openaire.eu/search/publication?articleId=doi_________::f5790aa4b4f43869882fa05bc6f92551Test https://doi.org/10.2337/db20-993-pTest |
| حقوق: | CLOSED |
| رقم الانضمام: | edsair.doi...........f5790aa4b4f43869882fa05bc6f92551 |
| قاعدة البيانات: | OpenAIRE |
| تدمد: | 1939327X 00121797 |
|---|