Introduction: Psychological stress may adversely affect glucose control. However, the effect of stress on efficacy of automated insulin delivery (AID) in Type 1 Diabetes (T1D) is still unknown. We evaluated non-glucose signals in subjects with type 1 diabetes on AID using the Empatica E4 device (E4) during psychological stress.

Method: During a 2-week prospective AID study for 12 participants with T1D, psychological stress i.e Trier Social Stress Test (TSST) was induced. Heart rate (HR) measurements using the Nonin sensor 15 mins before and after TSST confirmed psychological stress. Physical activity (PA) was analyzed and HR fluctuations coinciding with PA was rejected. The correlation of HR from E4 with the Nonin HR monitor was evaluated to explore the feasibility of using E4 HR as a substitute for Nonin HR. Pearson's moment is used to evaluate correlation.

Results: The HR from both Nonin and E4 was observed to rise after application of stress from baseline levels. The mean difference of heart rate in Nonin HR was 13.1 bpm (6 bpm) in AID and in E4 was 10.8 bpm (5 bpm) in AID. The Pearson's correlation coefficient between the Nonin and E4 HR was 0.9 in AID (p = 0.0007) . There was a trend towards increase in E4 Electrodermal activity (EDA) of normalized power spectral density (PSD) -92.5 dB/Hz (28.8) before stress to -84.9 dB/Hz (21.9) after stress in AID (p = 0.06) .

Conclusions: The E4 HR can act as a substitute to the gold standard Nonin HR. E4 HR is a good indicator of psychological stress with simultaneous analyses of accelerometry data and could be used to inform the AID controller. E4 EDA PSD is also a good indicator for stress but has a delayed response. On the other hand, HR can be affected by physical activity. Psychological stress could be detected through a fusion of E4 HR, EDA, and accelerometer sensors and incorporated into Next Gen AID for testing.


A.Banerjee: None. R.Kaur: None. I.Zaniletti: None. M.Church: None. S.K.Mccrady-spitzer: None. D.Desjardins: None. S.Gupta: None. Y.C.Kudva: Advisory Panel; Novo Nordisk, Research Support; Dexcom, Inc., Hemsley Charitable Trust, JDRF, National Institutes of Health, Roche Diabetes Care, Tandem Diabetes Care, Inc.


JDRF grant (2-SRA-2017-503-M-B)

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