Assessing Cognitive Workload in Motor Decision-Making through Functional Connectivity Analysis: Towards Early Detection and Monitoring of Neurodegenerative Diseases

Abstract

Neurodegenerative diseases (NDs), such as Alzheimer's, Parkinson's, amyotrophic lateral sclerosis, and frontotemporal dementia, among others, are increasingly prevalent in the global population. The clinical diagnosis of these NDs is based on the detection and characterization of motor and non-motor symptoms. However, when these diagnoses are made, the subjects are often in advanced stages where neuromuscular alterations are frequently irreversible. In this context, we propose a methodology to evaluate the cognitive workload (CWL) of motor tasks involving decision-making processes. CWL is a concept widely used to address the balance between task demand and the subject's available resources to complete that task. In this study, multiple models for motor planning during a motor decision-making task were developed by recording EEG and EMG signals in n=17 healthy volunteers (9 males, 8 females, age 28.66±8.8 years). In the proposed test, volunteers have to make decisions about which hand should be moved based on the onset of a visual stimulus. We computed functional connectivity between the cortex and muscles, as well as among muscles using both corticomuscular and intermuscular coherence. Despite three models being generated, just one of them had strong performance. The results showed two types of motor decision-making processes depending on the hand to move. Moreover, the central processing of decision-making for the left hand movement can be accurately estimated using behavioral measures such as planning time combined with peripheral recordings like EMG signals. The models provided in this study could be considered as a methodological foundation to detect neuromuscular alterations in asymptomatic patients, as well as to monitor the process of a degenerative disease.

Keywords: cognitive workload; decision-making; functional connectivity; motor planning; neurodegenerative diseases; statistical modeling.

Figure 1

(A) Experimental setup. (B) Variables’ interaction for modeling. (C) Sixty-four-channel EEG configuration system; electrodes in blue represent the premotor area.

Figure 2

Modeling procedure flowchart. In the upper part, each step of the modeling procedure. In the lower part, three-dimensional representations of the resampled data: for the bilateral model (black dots), for the right model (green dots), for the left model (red dots). The colored planes represent the developed models.

Figure 3

Distribution of resampled data using the bootstrap procedure. The original data are represented by circles (o), while the resampled data are represented by dots (·). In the left column, ×100 resampling; in the right column, ×1000 resampling. In the top row, data selection for the bilateral model (in black); in the middle row, data for the right hand model (in green); in the bottom row, data for the left hand model (in red).

Figure 4

Distribution of resampled data according to the connectivity variables. The shaded rectangle (in yellow) represents the quadrant formed by positive values for the corticomuscular connectivity variable ($C{C}_{CMC}$) and negative values for the intermuscular connectivity variable ($C{C}_{IMC}$). In the left figure, resampled data for the bilateral model are presented (in black), with 95% of the data in the expected quadrant. In the middle figure, resampled data for the right hand model are shown (in green), with 80% of the data in the expected quadrant. In the right figure, resampled data for the left hand model are displayed (in red), with 85% of the data in the expected quadrant.

Figure 5

Scatter plot of model fit. The X-axis represents the values of the original data for $C{C}_{CMC}$; the Y-axis represents the model outputs. In the left figure, the fit distribution for $M_{BILATERAL}$ is depicted in black; in the middle figure, $M_{RIGHT}$ is represented in green; in the right figure, $M_{LEFT}$ is shown in red. The solid lines (-) represent the fitted line; dashed and dotted lines (-·-) represent the limits of the standard error of estimation (SEE).

Figure 6

Graphical representation of the left hand motor planning model. (A) The right (contralateral) premotor area synchronizes with the left anterior deltoid muscle (agonist muscle); both anterior deltoid muscles desynchronize between them; the motor planning time is shorter for the left hand compared to the right hand. (B) Strong correlation (0.84) between connectivity changes. Moderate and inverse correlation between corticomuscular synchronization (−0.41) and planning time, as well as between intermuscular desynchronization (−0.61) and planning time. (C) Three-dimensional representation of the fitting plane of the left hand model ($M_{LEFT}$), showing the distribution trend and high predictive capacity. * p < 0.01 ** p < 0.05.