A Non-parametric Approach to the Overall Estimate of Cognitive Load Using NIRS Time Series

Abstract

We present a non-parametric approach to prediction of the n-back n ∈ {1, 2} task as a proxy measure of mental workload using Near Infrared Spectroscopy (NIRS) data. In particular, we focus on measuring the mental workload through hemodynamic responses in the brain induced by these tasks, thereby realizing the potential that they can offer for their detection in real world scenarios (e.g., difficulty of a conversation). Our approach takes advantage of intrinsic linearity that is inherent in the components of the NIRS time series to adopt a one-step regression strategy. We demonstrate the correctness of our approach through its mathematical analysis. Furthermore, we study the performance of our model in an inter-subject setting in contrast with state-of-the-art techniques in the literature to show a significant improvement on prediction of these tasks (82.50 and 86.40% for female and male participants, respectively). Moreover, our empirical analysis suggest a gender difference effect on the performance of the classifiers (with male data exhibiting a higher non-linearity) along with the left-lateralized activation in both genders with higher specificity in females.

Differential Effect of the Physical Embodiment on the Prefrontal Cortex Activity as Quantified by Its Entropy

“…The aforementioned findings are in line with the concept of entropy in information theory [ 86 ]. These results support the utility of entropy to the solution concept of the decoding of the information content of physiological and neurophysiological signals [ 87 , 88 ]. Extending these empirical results, Keshmiri et al [ 64 ] proved that differential entropy (DE, i.e., entropy of a continuous random variable) quantifies the information content of the frontal brain activity in terms of a shift (e.g., increase and/or decrease) in its variational information.…”

Differential Entropy Preserves Variational Information of Near-Infrared Spectroscopy Time Series Associated With Working Memory

“…In this respect, Proposition 0.0.2 implied that the adaptation of the slope of the NIRS time series of brain activity results in loss of such variational information. On the other hand, Keshmiri et al ( 2017 ) suggest the ability of DE of NIRS time series in outperforming the averaging-based feature spaces for decoding of the brain activity during N-Back working memory task. Whereas their results provide evidence on utility of DE in supervised learning paradigms, it is of significant importance to examine its ability in preservation of the information content of the brain activity in tasks with the underlying cognitive load of whose exhibit higher degree of non-triviality due to higher subjectivity of responses of human subjects (e.g., naturalistic scenarios such as conversation and story comprehension).…”

“…Shi et al ( 2013 ) suggest the applicability of differential entropy (DE) i.e., the entropy of a continuous random variable (Cover and Thomas, 2006 ) in analysis of electroencephalography (EEG) time series. Furthermore, Keshmiri et al ( 2017 ) show that DE significantly improves the classification accuracy of NIRS time series pertinent to cognitive load in prefrontal cortex during working memory task in comparison with a number of feature extraction and classification strategies. These results suggest the utility of DE to the solution concept of decoding of physiological and neurophysiological signals.…”

Entropy and the Brain: An Overview

“…For instance, Zheng and Lu [ 32 ] showed that the use of entropy for emotion classification based on electroencephalography (EEG) time series outperformed such feature spaces as differential asymmetry (DASM), rational asymmetry (RASM), and power spectral density (PSD). Similarly, Keshmiri et al [ 33 ] verified the discriminative power of entropy in comparison with the dominant feature spaces in near-infrared spectroscopy (fNIRS) analysis of the n-back WM task [ 34 ]. Most recently, Liu and colleagues [ 35 ] used a considerably large resting-state functional magnetic resonance imaging dataset from the Human Connectome Project (998 individuals).…”