多变量人工胰腺，适用于各种运动类型和强度。
Multivariable Artificial Pancreas for Various Exercise Types and Intensities.
机构信息
1 Department of Biomedical Engineering and Illinois Institute of Technology , Chicago, Illinois.
2 Department of Chemical and Biological Engineering, Illinois Institute of Technology , Chicago, Illinois.
出版信息
Diabetes Technol Ther. 2018 Oct;20(10):662-671. doi: 10.1089/dia.2018.0072. Epub 2018 Sep 6.
BACKGROUND
Exercise challenges people with type 1 diabetes in controlling their glucose concentration (GC). A multivariable adaptive artificial pancreas (MAAP) may lessen the burden.
METHODS
The MAAP operates without any user input and computes insulin based on continuous glucose monitor and physical activity signals. To analyze performance, 18 60-h closed-loop experiments with 96 exercise sessions with three different protocols were completed. Each day, the subjects completed one resistance and one treadmill exercise (moderate continuous training [MCT] or high-intensity interval training [HIIT]). The primary outcome is time spent in each glycemic range during the exercise + recovery period. Secondary measures include average GC and average change in GC during each exercise modality.
RESULTS
The GC during exercise + recovery periods were within the euglycemic range (70-180 mg/dL) for 69.9% of the time and within a safe glycemic range for exercise (70-250 mg/dL) for 93.0% of the time. The exercise sessions are defined to begin 30 min before the start of exercise and end 2 h after start of exercise. The GC were within the severe hypoglycemia (<55 mg/dL), moderate hypoglycemia (55-70 mg/dL), moderate hyperglycemia (180-250 mg/dL), and severe hyperglycemia (>250 mg/dL) for 0.9%, 1.3%, 23.1%, and 4.8% of the time, respectively. The average GC decline during exercise differed with exercise type (P = 0.0097) with a significant difference between the MCT and resistance (P = 0.0075). To prevent large GC decreases leading to hypoglycemia, MAAP recommended carbohydrates in 59% of MCT, 50% of HIIT, and 39% of resistance sessions.
CONCLUSIONS
A consistent GC decline occurred in exercise and recovery periods, which differed with exercise type. The average GC at the start of exercise was above target (185.5 ± 56.6 mg/dL for MCT, 166.9 ± 61.9 mg/dL for resistance training, and 171.7 ± 41.4 mg/dL HIIT), making a small decrease desirable. Hypoglycemic events occurred in 14.6% of exercise sessions and represented only 2.22% of the exercise and recovery period.
背景
运动给 1 型糖尿病患者的血糖浓度（GC）控制带来挑战。多变量自适应人工胰腺（MAAP）可能会减轻这种负担。
方法
MAAP 无需任何用户输入即可运行，并根据连续血糖监测仪和身体活动信号计算胰岛素。为了分析性能，完成了 18 个 60 小时的闭环实验，其中有 96 个运动疗程，采用了三种不同的方案。每天，受试者完成一次抗阻运动和一次跑步机运动（中等持续训练[MCT]或高强度间歇训练[HIIT]）。主要结果是运动和恢复期内每个血糖范围的时间。次要指标包括每种运动模式下的平均 GC 和平均 GC 变化。
结果
运动和恢复期内的 GC 有 69.9%的时间处于正常血糖范围（70-180mg/dL），有 93.0%的时间处于运动安全血糖范围（70-250mg/dL）。运动疗程定义为运动开始前 30 分钟开始，运动开始后 2 小时结束。GC 分别有 0.9%、1.3%、23.1%和 4.8%的时间处于严重低血糖（<55mg/dL）、中度低血糖（55-70mg/dL）、中度高血糖（180-250mg/dL）和严重高血糖（>250mg/dL）。运动类型不同，GC 平均下降也不同（P=0.0097），MCT 和抗阻运动之间存在显著差异（P=0.0075）。为了防止 GC 大幅下降导致低血糖，MAAP 建议在 59%的 MCT、50%的 HIIT 和 39%的抗阻运动中摄入碳水化合物。
结论
运动和恢复期内的 GC 持续下降，与运动类型不同。运动开始时的平均 GC 高于目标值（MCT 为 185.5±56.6mg/dL，抗阻训练为 166.9±61.9mg/dL，HIIT 为 171.7±41.4mg/dL），因此稍有下降是可取的。低血糖事件发生在 14.6%的运动疗程中，仅占运动和恢复期的 2.22%。
Zotero 插件