Autonomic modulation during a cognitive task using a wearable device

Abstract

Heart-brain interaction is by nature bidirectional, and then, it is sensible to expect the heart, via the autonomic nervous system (ANS), to induce changes in the brain. Respiration can originate differentiated ANS states reflected by HRV. In this work, we measured the changes in performance during a cognitive task due to four autonomic states originated by breath control: at normal breathing (NB), fast breathing (FB), slow breathing (SB) and control phases. ANS states were characterized by temporal (SDNN) and spectral (LF and HF power) HRV markers. Cognitive performance was measured by the response time (RT) and the success rate (SR). HRV parameters were acquired with the wristband Empatica E4. Classification was accomplished, firstly, to find the best ANS variables that discriminated the breathing phases (BPH) and secondly, to find whether ANS parameters were associated to changes in RT and SR. In order to compensate for possible bias of the test sets, 1000 classification iterations were run. The ANS parameters that better separated the four BPH were LF and HF power, with changes about 300% from controls and an average classification rate of 59.9%, a 34.9% more than random. LF and HF explained RT separation for every BPH pair, and so was HF for SR separation. The best RT classification was 63.88% at NB vs SB phases, while SR provided a 73.39% at SB vs NB phases. Results suggest that breath control could show a relation with the efficiency of certain cognitive tasks. For this goal the Empatica wristband together with the proposed methodology could help to clarify this hypothesis.