EEG-derived markers to improve prognostic evaluation of Disorders of Consciousness.

Disorders of consciousness (DoC) are characterized by alteration in arousal and/or awareness commonly caused by severe brain injury. There exists a consensus on adopting advanced neuroimaging and electrophysiological procedures to improve diagnosis/prognosis of DoC patients. Currently, these procedures are prevalently applied in a research-oriented context and their translation into clinical practice is yet to come. The aim of the study consisted in the identification of measures derived from routinary electroencephalography (EEG) able to support clinicians in the prediction of DoC patients' outcome. In the present study, a routine EEG was recorded during rest from a sample of 58 DoC patients clinically diagnosed as Unresponsive Wakefulness State (UWS) and Minimally Conscious State (MCS) and followed- up for 3 months. EEG-based features characterizing brain activity in terms of spectral content and resting state networks organization were used in a predictive machine learning model to i) identify which were the most promising features in predicting patients' exit from the DoC, regardless of the clinical diagnosis and ii) verify whether such features would have been the same best discriminating UWS from MCS or specific of the outcome prediction. A predictive machine learning model was built on EEG features related to spectral content and resting state networks which returned up to 85% of performance accuracy in outcome prediction and 76% in DoC state recognition (UWS vs MCS). We provided preliminary evidence for the exploitation of a routine EEG to improve the clinical management of non- communicative patients to be confirmed in a larger DoC population.

The Promotoer, a brain-computer interface-assisted intervention to promote upper limb functional motor recovery after stroke: a statistical analysis plan for a randomized controlled trial.

BACKGROUND: Electroencephalography (EEG)-based brain-computer interfaces (BCIs) allow to modulate the sensorimotor rhythms and are emerging technologies for promoting post-stroke motor function recovery. The Promotoer study aims to assess the short and long-term efficacy of the Promotoer system, an EEG-based BCI assisting motor imagery (MI) practice, in enhancing post-stroke functional hand motor recovery. This paper details the statistical analysis plan of the Promotoer study. METHODS: The Promotoer study is a randomized, controlled, assessor-blinded, single-centre, superiority trial, with two parallel groups and a 1:1 allocation ratio. Subacute stroke patients are randomized to EEG-based BCI-assisted MI training or to MI training alone (i.e. no BCI). An internal pilot study for sample size re-assessment is planned. The primary outcome is the effectiveness of the Upper Extremity Fugl-Meyer Assessment (UE-FMA) score. Secondary outcomes include clinical, functional, and user experience scores assessed at the end of intervention and at follow-up. Neurophysiological assessments are also planned. Effectiveness formulas have been specified, and intention-to-treat and per-protocol populations have been defined. Statistical methods for comparisons of groups and for development of a predictive score of significant improvement are described. Explorative subgroup analyses and methodology to handle missing data are considered. DISCUSSION: The Promotoer study will provide robust evidence for the short/long-term efficacy of the Promotoer system in subacute stroke patients undergoing a rehabilitation program. Moreover, the development of a predictive score of response will allow transferring of the Promotoer system to optimal clinical practice. By carefully describing the statistical principles and procedures, the statistical analysis plan provides transparency in the analysis of data. TRIAL REGISTRATION: ClinicalTrials.gov NCT04353297 . Registered on April 15, 2020.

DiSCIoser: unlocking recovery potential of arm sensorimotor functions after spinal cord injury by promoting activity-dependent brain plasticity by means of brain-computer interface technology: a randomized controlled trial to test efficacy.

BACKGROUND: Traumatic cervical spinal cord injury (SCI) results in reduced sensorimotor abilities that strongly impact on the achievement of daily living activities involving hand/arm function. Among several technology-based rehabilitative approaches, Brain-Computer Interfaces (BCIs) which enable the modulation of electroencephalographic sensorimotor rhythms, are promising tools to promote the recovery of hand function after SCI. The "DiSCIoser" study proposes a BCI-supported motor imagery (MI) training to engage the sensorimotor system and thus facilitate the neuroplasticity to eventually optimize upper limb sensorimotor functional recovery in patients with SCI during the subacute phase, at the peak of brain and spinal plasticity. To this purpose, we have designed a BCI system fully compatible with a clinical setting whose efficacy in improving hand sensorimotor function outcomes in patients with traumatic cervical SCI will be assessed and compared to the hand MI training not supported by BCI. METHODS: This randomized controlled trial will include 30 participants with traumatic cervical SCI in the subacute phase randomly assigned to 2 intervention groups: the BCI-assisted hand MI training and the hand MI training not supported by BCI. Both interventions are delivered (3 weekly sessions; 12 weeks) as add-on to standard rehabilitation care. A multidimensional assessment will be performed at: randomization/pre-intervention and post-intervention. Primary outcome measure is the Graded Redefined Assessment of Strength, Sensibility and Prehension (GRASSP) somatosensory sub-score. Secondary outcome measures include the motor and functional scores of the GRASSP and other clinical, neuropsychological, neurophysiological and neuroimaging measures. DISCUSSION: We expect the BCI-based intervention to promote meaningful cortical sensorimotor plasticity and eventually maximize recovery of arm functions in traumatic cervical subacute SCI. This study will generate a body of knowledge that is fundamental to drive optimization of BCI application in SCI as a top-down therapeutic intervention, thus beyond the canonical use of BCI as assistive tool. TRIAL REGISTRATION: Name of registry: DiSCIoser: improving arm sensorimotor functions after spinal cord injury via brain-computer interface training (DiSCIoser). TRIAL REGISTRATION NUMBER: NCT05637775; registration date on the ClinicalTrial.gov platform: 05-12-2022.

Parallel Factorization to Implement Group Analysis in Brain Networks Estimation.

When dealing with complex functional brain networks, group analysis still represents an open issue. In this paper, we investigated the potential of an innovative approach based on PARAllel FActorization (PARAFAC) for the extraction of the grand average connectivity matrices from both simulated and real datasets. The PARAFAC approach was solved using three different numbers of rank-one tensors (PAR-FACT). Synthetic data were parametrized according to different levels of three parameters: network dimension (NODES), number of observations (SAMPLE-SIZE), and noise (SWAP-CON) in order to investigate the way they affect the grand average estimation. PARAFAC was then tested on a real connectivity dataset, derived from EEG data of 17 healthy subjects performing wrist extension with left and right hand separately. Findings on both synthetic and real data revealed the potential of the PARAFAC algorithm as a useful tool for grand average extraction. As expected, the best performances in terms of FPR, FNR, and AUC were achieved for great values of sample size and low noise level. A crucial role has been revealed for the PAR-FACT parameter, revealing that an increase in the number of rank-one tensors solving the PARAFAC problem leads to an increase in FPR values and, thus, to a worse grand average estimation.

Exploring high-density corticomuscular networks after stroke to enable a hybrid Brain-Computer Interface for hand motor rehabilitation.

BACKGROUND: Brain-Computer Interfaces (BCI) promote upper limb recovery in stroke patients reinforcing motor related brain activity (from electroencephalogaphy, EEG). Hybrid BCIs which include peripheral signals (electromyography, EMG) as control features could be employed to monitor post-stroke motor abnormalities. To ground the use of corticomuscular coherence (CMC) as a hybrid feature for a rehabilitative BCI, we analyzed high-density CMC networks (derived from multiple EEG and EMG channels) and their relation with upper limb motor deficit by comparing data from stroke patients with healthy participants during simple hand tasks. METHODS: EEG (61 sensors) and EMG (8 muscles per arm) were simultaneously recorded from 12 stroke (EXP) and 12 healthy participants (CTRL) during simple hand movements performed with right/left (CTRL) and unaffected/affected hand (EXP, UH/AH). CMC networks were estimated for each movement and their properties were analyzed by means of indices derived ad-hoc from graph theory and compared among groups. RESULTS: Between-group analysis showed that CMC weight of the whole brain network was significantly reduced in patients during AH movements. The network density was increased especially for those connections entailing bilateral non-target muscles. Such reduced muscle-specificity observed in patients was confirmed by muscle degree index (connections per muscle) which indicated a connections' distribution among non-target and contralateral muscles and revealed a higher involvement of proximal muscles in patients. CMC network properties correlated with upper-limb motor impairment as assessed by Fugl-Meyer Assessment and Manual Muscle Test in patients. CONCLUSIONS: High-density CMC networks can capture motor abnormalities in stroke patients during simple hand movements. Correlations with upper limb motor impairment support their use in a BCI-based rehabilitative approach.

Cortico-muscular coupling to control a hybrid brain-computer interface for upper limb motor rehabilitation: A pseudo-online study on stroke patients.

Brain-Computer Interface (BCI) systems for motor rehabilitation after stroke have proven their efficacy to enhance upper limb motor recovery by reinforcing motor related brain activity. Hybrid BCIs (h-BCIs) exploit both central and peripheral activation and are frequently used in assistive BCIs to improve classification performances. However, in a rehabilitative context, brain and muscular features should be extracted to promote a favorable motor outcome, reinforcing not only the volitional control in the central motor system, but also the effective projection of motor commands to target muscles, i.e., central-to-peripheral communication. For this reason, we considered cortico-muscular coupling (CMC) as a feature for a h-BCI devoted to post-stroke upper limb motor rehabilitation. In this study, we performed a pseudo-online analysis on 13 healthy participants (CTRL) and 12 stroke patients (EXP) during executed (CTRL, EXP unaffected arm) and attempted (EXP affected arm) hand grasping and extension to optimize the translation of CMC computation and CMC-based movement detection from offline to online. Results showed that updating the CMC computation every 125 ms (shift of the sliding window) and accumulating two predictions before a final classification decision were the best trade-off between accuracy and speed in movement classification, independently from the movement type. The pseudo-online analysis on stroke participants revealed that both attempted and executed grasping/extension can be classified through a CMC-based movement detection with high performances in terms of classification speed (mean delay between movement detection and EMG onset around 580 ms) and accuracy (hit rate around 85%). The results obtained by means of this analysis will ground the design of a novel non-invasive h-BCI in which the control feature is derived from a combined EEG and EMG connectivity pattern estimated during upper limb movement attempts.

Automatic Selection of Control Features for Electroencephalography-Based Brain-Computer Interface Assisted Motor Rehabilitation: The GUIDER Algorithm.

Sensorimotor rhythms-based Brain-Computer Interfaces (BCIs) have successfully been employed to address upper limb motor rehabilitation after stroke. In this context, becomes crucial the choice of features that would enable an appropriate electroencephalographic (EEG) sensorimotor activation/engagement underlying the favourable motor recovery. Here, we present a novel feature selection algorithm (GUIDER) designed and implemented to integrate specific requirements related to neurophysiological knowledge and rehabilitative principles. The GUIDER algorithm was tested on an EEG dataset collected from 13 subacute stroke participants. The comparison between the automatic feature selection procedure by means of GUIDER algorithm and the manual feature selection executed by an expert neurophysiologist returned similar performance in terms of both feature selection and classification. Our preliminary findings suggest that the choices of experienced neurophysiologists could be reproducible by an automatic approach. The proposed automatic algorithm could be apt to support the professional end-users not expert in BCI such as therapist/clinicians and, to ultimately foster a wider employment of the BCI-based rehabilitation after stroke.

Risk factors for 2-year mortality in patients with prolonged disorders of consciousness: An international multicentre study.

BACKGROUND AND PURPOSE: Patients with prolonged disorders of consciousness (pDoC) have a high mortality rate due to medical complications. Because an accurate prognosis is essential for decision-making on patients' management, we analysed data from an international multicentre prospective cohort study to evaluate 2-year mortality rate and bedside predictors of mortality. METHODS: We enrolled adult patients in prolonged vegetative state/unresponsive wakefulness syndrome (VS/UWS) or minimally conscious state (MCS) after traumatic and nontraumatic brain injury within 3 months postinjury. At enrolment, we collected demographic (age, sex), anamnestic (aetiology, time postinjury), clinical (Coma Recovery Scale-Revised [CRS-R], Disability Rating Scale, Nociception Coma Scale-Revised), and neurophysiologic (electroencephalogram [EEG], somatosensory evoked and event-related potentials) data. Patients were followed up to gather data on mortality up to 24 months postinjury. RESULTS: Among 143 traumatic (n = 55) and nontraumatic (n = 88) patients (VS/UWS, n = 68, 19 females; MCS, n = 75, 22 females), 41 (28.7%) died within 24 months postinjury. Mortality rate was higher in VS/UWS (42.6%) than in MCS (16%; p < 0.001). Multivariate regression in VS/UWS showed that significant predictors of mortality were older age and lower CRS-R total score, whereas in MCS female sex and absence of alpha rhythm on EEG at study entry were significant predictors. CONCLUSIONS: This study demonstrated that a feasible multimodal assessment in the postacute phase can help clinicians to identify patients with pDoC at higher risk of mortality within 24 months after brain injury. This evidence can help clinicians and patients' families to navigate the complex clinical decision-making process and promote an international standardization of prognostic procedures for patients with pDoC.

Corticomuscular and Intermuscular Coupling in Simple Hand Movements to Enable a Hybrid Brain-Computer Interface.

Hybrid Brain-Computer Interfaces (BCIs) for upper limb rehabilitation after stroke should enable the reinforcement of "more normal" brain and muscular activity. Here, we propose the combination of corticomuscular coherence (CMC) and intermuscular coherence (IMC) as control features for a novel hybrid BCI for rehabilitation purposes. Multiple electroencephalographic (EEG) signals and surface electromyography (EMG) from 5 muscles per side were collected in 20 healthy participants performing finger extension (Ext) and grasping (Grasp) with both dominant and non-dominant hand. Grand average of CMC and IMC patterns showed a bilateral sensorimotor area as well as multiple muscles involvement. CMC and IMC values were used as features to classify each task versus rest and Ext versus Grasp. We demonstrated that a combination of CMC and IMC features allows for classification of both movements versus rest with better performance (Area Under the receiver operating characteristic Curve, AUC) for the Ext movement (0.97) with respect to Grasp (0.88). Classification of Ext versus Grasp also showed high performances (0.99). All in all, these preliminary findings indicate that the combination of CMC and IMC could provide for a comprehensive framework for simple hand movements to eventually be employed in a hybrid BCI system for post-stroke rehabilitation.

SEED-G: Simulated EEG Data Generator for Testing Connectivity Algorithms.

EEG signals are widely used to estimate brain circuits associated with specific tasks and cognitive processes. The testing of connectivity estimators is still an open issue because of the lack of a ground-truth in real data. Existing solutions such as the generation of simulated data based on a manually imposed connectivity pattern or mass oscillators can model only a few real cases with limited number of signals and spectral properties that do not reflect those of real brain activity. Furthermore, the generation of time series reproducing non-ideal and non-stationary ground-truth models is still missing. In this work, we present the SEED-G toolbox for the generation of pseudo-EEG data with imposed connectivity patterns, overcoming the existing limitations and enabling control of several parameters for data simulation according to the user's needs. We first described the toolbox including guidelines for its correct use and then we tested its performances showing how, in a wide range of conditions, datasets composed by up to 60 time series were successfully generated in less than 5 s and with spectral features similar to real data. Then, SEED-G is employed for studying the effect of inter-trial variability Partial Directed Coherence (PDC) estimates, confirming its robustness.

Estimation of Granger causality through Artificial Neural Networks: applications to physiological systems and chaotic electronic oscillators.

One of the most challenging problems in the study of complex dynamical systems is to find the statistical interdependencies among the system components. Granger causality (GC) represents one of the most employed approaches, based on modeling the system dynamics with a linear vector autoregressive (VAR) model and on evaluating the information flow between two processes in terms of prediction error variances. In its most advanced setting, GC analysis is performed through a state-space (SS) representation of the VAR model that allows to compute both conditional and unconditional forms of GC by solving only one regression problem. While this problem is typically solved through Ordinary Least Square (OLS) estimation, a viable alternative is to use Artificial Neural Networks (ANNs) implemented in a simple structure with one input and one output layer and trained in a way such that the weights matrix corresponds to the matrix of VAR parameters. In this work, we introduce an ANN combined with SS models for the computation of GC. The ANN is trained through the Stochastic Gradient Descent L1 (SGD-L1) algorithm, and a cumulative penalty inspired from penalized regression is applied to the network weights to encourage sparsity. Simulating networks of coupled Gaussian systems, we show how the combination of ANNs and SGD-L1 allows to mitigate the strong reduction in accuracy of OLS identification in settings of low ratio between number of time series points and of VAR parameters. We also report how the performances in GC estimation are influenced by the number of iterations of gradient descent and by the learning rate used for training the ANN. We recommend using some specific combinations for these parameters to optimize the performance of GC estimation. Then, the performances of ANN and OLS are compared in terms of GC magnitude and statistical significance to highlight the potential of the new approach to reconstruct causal coupling strength and network topology even in challenging conditions of data paucity. The results highlight the importance of of a proper selection of regularization parameter which determines the degree of sparsity in the estimated network. Furthermore, we apply the two approaches to real data scenarios, to study the physiological network of brain and peripheral interactions in humans under different conditions of rest and mental stress, and the effects of the newly emerged concept of remote synchronization on the information exchanged in a ring of electronic oscillators. The results highlight how ANNs provide a mesoscopic description of the information exchanged in networks of multiple interacting physiological systems, preserving the most active causal interactions between cardiovascular, respiratory and brain systems. Moreover, ANNs can reconstruct the flow of directed information in a ring of oscillators whose statistical properties can be related to those of physiological networks.

Multicenter prospective study on predictors of short-term outcome in disorders of consciousness.

OBJECTIVE: This international multicenter, prospective, observational study aimed at identifying predictors of short-term clinical outcome in patients with prolonged disorders of consciousness (DoC) due to acquired severe brain injury. METHODS: Patients in vegetative state/unresponsive wakefulness syndrome (VS/UWS) or in minimally conscious state (MCS) were enrolled within 3 months from their brain injury in 12 specialized medical institutions. Demographic, anamnestic, clinical, and neurophysiologic data were collected at study entry. Patients were then followed up for assessing the primary outcome, that is, clinical diagnosis according to standardized criteria at 6 months postinjury. RESULTS: We enrolled 147 patients (44 women; mean age 49.4 [95% confidence interval 46.1-52.6] years; VS/UWS 71, MCS 76; traumatic 55, vascular 56, anoxic 36; mean time postinjury 59.6 [55.4-63.6] days). The 6-month follow-up was complete for 143 patients (VS/UWS 70; MCS 73). With respect to study entry, the clinical diagnosis improved in 72 patients (VS/UWS 27; MCS 45). Younger age, shorter time postinjury, higher Coma Recovery Scale-Revised total score, and presence of EEG reactivity to eye opening at study entry predicted better outcome, whereas etiology, clinical diagnosis, Disability Rating Scale score, EEG background activity, acoustic reactivity, and P300 on event-related potentials were not associated with outcome. CONCLUSIONS: Multimodal assessment could identify patients with higher likelihood of clinical improvement in order to help clinicians, families, and funding sources with various aspects of decision-making. This multicenter, international study aims to stimulate further research that drives international consensus regarding standardization of prognostic procedures for patients with DoC.

The Promotoer, a brain-computer interface-assisted intervention to promote upper limb functional motor recovery after stroke: a study protocol for a randomized controlled trial to test early and long-term efficacy and to identify determinants of response.

BACKGROUND: Stroke is a leading cause of long-term disability. Cost-effective post-stroke rehabilitation programs for upper limb are critically needed. Brain-Computer Interfaces (BCIs) which enable the modulation of Electroencephalography (EEG) sensorimotor rhythms are promising tools to promote post-stroke recovery of upper limb motor function. The "Promotoer" study intends to boost the application of the EEG-based BCIs in clinical practice providing evidence for a short/long-term efficacy in enhancing post-stroke hand functional motor recovery and quantifiable indices of the participants response to a BCI-based intervention. To these aims, a longitudinal study will be performed in which subacute stroke participants will undergo a hand motor imagery (MI) training assisted by the Promotoer system, an EEG-based BCI system fully compliant with rehabilitation requirements. METHODS: This longitudinal 2-arm randomized controlled superiority trial will include 48 first ever, unilateral, subacute stroke participants, randomly assigned to 2 intervention groups: the BCI-assisted hand MI training and a hand MI training not supported by BCI. Both interventions are delivered (3 weekly session; 6 weeks) as add-on regimen to standard intensive rehabilitation. A multidimensional assessment will be performed at: randomization/pre-intervention, 48 h post-intervention, and at 1, 3 and 6 month/s after end of intervention. Primary outcome measure is the Fugl-Meyer Assessment (FMA, upper extremity) at 48 h post-intervention. Secondary outcome measures include: the upper extremity FMA at follow-up, the Modified Ashworth Scale, the Numeric Rating Scale for pain, the Action Research Arm Test, the National Institute of Health Stroke Scale, the Manual Muscle Test, all collected at the different timepoints as well as neurophysiological and neuroimaging measures. DISCUSSION: We expect the BCI-based rewarding of hand MI practice to promote long-lasting retention of the early induced improvement in hand motor outcome and also, this clinical improvement to be sustained by a long-lasting neuroplasticity changes harnessed by the BCI-based intervention. Furthermore, the longitudinal multidimensional assessment will address the selection of those stroke participants who best benefit of a BCI-assisted therapy, consistently advancing the transfer of BCIs to a best clinical practice. TRIAL REGISTRATION: Name of registry: BCI-assisted MI Intervention in Subacute Stroke (Promotoer). TRIAL REGISTRATION NUMBER: NCT04353297 ; registration date on the ClinicalTrial.gov platform: April, 15/2020.

Raising the bar: Can dual scanning improve our understanding of joint action?

Two-person neuroscience (2 â€‹PN) is a recently introduced conceptual and methodological framework used to investigate the neural basis of human social interaction from simultaneous neuroimaging of two or more subjects (hyperscanning). In this study, we adopted a 2 â€‹PN approach and a multiple-brain connectivity model to investigate the neural basis of a form of cooperation called joint action. We hypothesized different intra-brain and inter-brain connectivity patterns when comparing the interpersonal properties of joint action with non-interpersonal conditions, with a focus on co-representation, a core ability at the basis of cooperation. 32 subjects were enrolled in dual-EEG recordings during a computerized joint action task including three conditions: one in which the dyad jointly acted to pursue a common goal (joint), one in which each subject interacted with the PC (PC), and one in which each subject performed the task individually (Solo). A combination of multiple-brain connectivity estimation and specific indices derived from graph theory allowed to compare interpersonal with non-interpersonal conditions in four different frequency bands. Our results indicate that all the indices were modulated by the interaction, and returned a significantly stronger integration of multiple-subject networks in the joint vs. PC and Solo conditions. A subsequent classification analysis showed that features based on multiple-brain indices led to a better discrimination between social and non-social conditions with respect to single-subject indices. Taken together, our results suggest that multiple-brain connectivity can provide a deeper insight into the understanding of the neural basis of cooperation in humans.

Language-Related Brain Potentials in Patients With Disorders of Consciousness: A Follow-up Study to Detect "Covert" Language Disorders.

Background. Language disorders may occur in patients with disorders of consciousness (DoCs), and they could interfere with the behavioral assessment of consciousness and responsiveness. Objective. In this study, we retrospectively explored whether ERP N400 was eventually associated with the presence of aphasia diagnosed in those patients who had evolved into Exit-Minimally Conscious State (E-MCS) at the clinical follow-up. Methods. In this retrospective cohort study, the ERPs elicited by an auditory sentences task were retrospectively examined in a sample of 15 DoC patients diagnosed according to the Coma Recovery Scale-Revised (CRS-R). All these 15 DoC patients underwent a (at least) 1-year clinical follow-up, which included a neuropsychological evaluation to assess language function among other cognitive functions. Ten healthy individuals also underwent the same paradigm to investigate the variability of ERPs characteristics. Results. The N400 ERP component with centroparietal topography was found in 9 of 10 healthy controls in response to the ill-formed sentences. Analysis of patients' data revealed that (1) a significant N400 component could be detected in 64% (9 of 14 patients) of the DoC patients regardless of the type of DoC; (2) no significant N400 ERP component was retrospectively detected in those E-MCS patients who showed aphasia at the follow-up; and (3) the presence/absence of the N400-ERP component was consistent with the brain lesion side and significantly predict the recovery. Conclusion. These preliminary findings indicate that the absence of N400 was significantly associated with the presence of aphasia diagnosed at the clinical follow-up in E-MCS patients.

Estimation of brain connectivity through Artificial Neural Networks.

Among different methods available for estimating brain connectivity from electroencephalographic signals (EEG), those based on MVAR models have proved to be flexible and accurate. They rely on the solution of linear equations that can be pursued through artificial neural networks (ANNs) used as MVAR model. However, when few data samples are available, there is a lack of accuracy in estimating MVAR parameters due to the collinearity between regressors. Moreover, the assessment procedure is also affected by the lack of data points. The mathematical solution to these problems is represented by penalized regression methods based on l1 norm, that can reduce collinearity by means of variable selection process. However, the direct application of l1 norm during the training of an ANN does not result in an efficient learning process. With the introduction of the stochastic gradient descent-L1 (SGD-L1) it is possible to apply l1 norm directly on the estimated weights in an efficient way. Even if ANNs has been used as MVAR model for brain connectivity estimation, the use of SGD-L1 algorithm has never been tested to this purpose when few data samples are available. In this work, we tested an approach based on ANNs and SGD-L1 on both surrogate and real EEG data. Our results show that ANNs can provide accurate brain connectivity estimation if trained with SGD-L1 algorithm even when few data samples are available.

Single-trial Connectivity Estimation through the Least Absolute Shrinkage and Selection Operator.

Methods based on the use of multivariate autoregressive models (MVAR) have proved to be an accurate tool for the estimation of functional links between the activity originated in different brain regions. A well-established method for the parameters estimation is the Ordinary Least Square (OLS) approach, followed by an assessment procedure that can be performed by means of Asymptotic Statistic (AS). However, the performances of both procedures are strongly influenced by the number of data samples available, thus limiting the conditions in which brain connectivity can be estimated. The aim of this paper is to introduce and test a regression method based on Least Absolute Shrinkage and Selection Operator (LASSO) to broaden the estimation of brain connectivity to those conditions in which current methods fail due to the limited data points available. We tested the performances of the LASSO regression in a simulation study under different levels of data points available, in comparison with a classical approach based on OLS and AS. Then, the two methods were applied to real electroencephalographic (EEG) signals, recorded during a motor imagery task. The simulation study and the application to real EEG data both indicated that LASSO regression provides better performances than the currently used methodologies for the estimation of brain connectivity when few data points are available. This work paves the way to the estimation and assessment of connectivity patterns with limited data amount and in on-line settings.

How the workload impacts on cognitive cooperation: A pilot study.

Cooperation degradation can be seen as one of the main causes of human errors. Poor cooperation could arise from aberrant mental processes, such as mental overload, that negatively affect the user's performance. Using different levels of difficulty in a cooperative task, we combined behavioural, subjective and neurophysiological data with the aim to i) quantify the mental workload under which the crew was operating, ii) evaluate the degree of their cooperation, and iii) assess the impact of the workload demands on the cooperation levels. The combination of such data showed that high workload demand impacted significantly on the performance, workload perception, and degree of cooperation.

Estimating brain connectivity when few data points are available: Perspectives and limitations.

Methods based on the use of multivariate autoregressive modeling (MVAR) have proved to be an accurate and flexible tool for the estimation of brain functional connectivity. The multivariate approach, however, implies the use of a model whose complexity (in terms of number of parameters) increases quadratically with the number of signals included in the problem. This can often lead to an underdetermined problem and to the condition of multicollinearity. The aim of this paper is to introduce and test an approach based on Ridge Regression combined with a modified version of the statistics usually adopted for these methods, to broaden the estimation of brain connectivity to those conditions in which current methods fail, due to the lack of enough data points. We tested the performances of this new approach, in comparison with the classical approach based on ordinary least squares (OLS), by means of a simulation study implementing different ground-truth networks, under different network sizes and different levels of data points. Simulation results showed that the new approach provides better performances, in terms of accuracy of the parameters estimation and false positives/false negatives rates, in all conditions related to a low data points/model dimension ratio, and may thus be exploited to estimate and validate estimated patterns at single-trial level or when short time data segments are available.

Different Topological Properties of EEG-Derived Networks Describe Working Memory Phases as Revealed by Graph Theoretical Analysis.

Several non-invasive imaging methods have contributed to shed light on the brain mechanisms underlying working memory (WM). The aim of the present study was to depict the topology of the relevant EEG-derived brain networks associated to distinct operations of WM function elicited by the Sternberg Item Recognition Task (SIRT) such as encoding, storage, and retrieval in healthy, middle age (46 ± 5 years) adults. High density EEG recordings were performed in 17 participants whilst attending a visual SIRT. Neural correlates of WM were assessed by means of a combination of EEG signal processing methods (i.e., time-varying connectivity estimation and graph theory), in order to extract synthetic descriptors of the complex networks underlying the encoding, storage, and retrieval phases of WM construct. The group analysis revealed that the encoding phase exhibited a significantly higher small-world topology of EEG networks with respect to storage and retrieval in all EEG frequency oscillations, thus indicating that during the encoding of items the global network organization could "optimally" promote the information flow between WM sub-networks. We also found that the magnitude of such configuration could predict subject behavioral performance when memory load increases as indicated by the negative correlation between Reaction Time and the local efficiency values estimated during the encoding in the alpha band in both 4 and 6 digits conditions. At the local scale, the values of the degree index which measures the degree of in- and out- information flow between scalp areas were found to specifically distinguish the hubs within the relevant sub-networks associated to each of the three different WM phases, according to the different role of the sub-network of regions in the different WM phases. Our findings indicate that the use of EEG-derived connectivity measures and their related topological indices might offer a reliable and yet affordable approach to monitor WM components and thus theoretically support the clinical assessment of cognitive functions in presence of WM decline/impairment, as it occurs after stroke.