The accuracy of blood glucose concentration (BGC) estimation affects insulin dosing decisions. Wristband biosignals (Empatica E4) are used to estimate physical activity type and intensity, and psychological stress to enhance BGC prediction accuracy. Three cases are compared: (1) BGC prediction by using only continuous glucose monitor (CGM) data, which considers only meal effects and plasma insulin concentration (PIC); (2) Incorporating physical activity information to complement CGM data; and (3) Incorporating both physical activity and psychological stress information to complement CGM data. Twenty-eight experiments with seven subjects with T1D are conducted including various physical activities and acute psychological stress inducements. The subjects wore a CGM, insulin pump and E4 wristband during the experiments. Physical activities included treadmill running, stationary bicycling, resistance exercises and activities of daily living. Acute psychological stress inducements included the Trier Social Stress test, playing competitive videogames, solving arithmetic problems, and daily stressful events. The overall mean absolute errors (%) were 10.42, 8.69. and 8.42 for Cases 1, 2, and 3, respectively. The results illustrate that incorporating estimates of psychological and psychological effects improve the BGC prediction accuracy by 17.18% (Freidman’s ANOVA test, p-value < 0.01).


M. Sevil: None. N. Hobbs: None. M. Rashid: None. I. Hajizadeh: None. M. Askari: None. M. Park: None. L.T. Quinn: None. A. Cinar: Research Support; Self; Dexcom, Inc., JDRF. Stock/Shareholder; Self; Dexcom, Inc., Eli Lilly and Company, Insulet Corporation, Medtronic, Novo Nordisk A/S, Tandem Diabetes Care.


National Institutes of Health (1DP3DK101075-01, 1DP3DK101077-01); JDRF (A18-0036-001)

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