Glycemic control during and after exercise is challenging. APs reduce %<70 and TIR, but do not eliminate exercise-induced hypoglycemia. Recent work has shown that insulin reduction 90min prior to exercise may be optimal. We present the initial testing of an exercise-informed AP, able to anticipate regular exercise by up to 2h without user input. Eighteen adults with T1D were enrolled in a randomized crossover trial: 4-weeks data collection (DC: SAP+step counter) followed by two 32h supervised hotel stays (SH). During DC, participants were asked to exercise 4 times/week for >30min, between 4-7pm. During SH, participants used either a naïve AP (CTL) or exercise informed AP (EX) and participated in 45min of moderate exercise at 17:30 on day 1. Fifteen subjects completed the study, age 42.2±13.6, HbA1c 6.61±1.02, with no system related AEs. During exercise, EX and CTL had similar TIR 88.2 vs. 85.2%, with a tendency to less time <80mgdL, 14.8 vs. 8.2%, following a reduction of insulin from 2.4 to 1.8 U over the preceding 2h. Post exercise EX and CTL had similar TIR (87.2 vs. 85.5%) but EX had less %<70 1.1% vs. 8.9% p=0.04,overnight control was excellent in both case TIR: 98.1 vs. 99.4%; %<70: 0.1 vs. 0.6%. During the sedentary late afternoon (5-7pm day 2) there was no difference in TIR: 89.6 vs. 86.0%. APs can learn from prior exercise behaviors and anticipate them, potentially leading to reduced glycemic risk, and a more automated system.


J. Garcia-Tirado: None. S.A. Brown: Research Support; Self; Dexcom, Inc., Insulet Corporation, Roche Diabetes Care, Tandem Diabetes Care, Tolerion, Inc. N. Laichuthai: None. H.E. Myers: None. M.D. Breton: Research Support; Self; Dexcom, Inc., Novo Nordisk A/S, Sanofi, Tandem Diabetes Care. Speaker’s Bureau; Self; Dexcom, Inc., Tandem Diabetes Care.


National Institutes of Health (DP3DK106826)

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at