Special Report on the Impact of the COVID-19 Pandemic on Clinical EEG and Research and Consensus Recommendations for the Safe Use of EEG.

INTRODUCTION: The global COVID-19 pandemic has affected the economy, daily life, and mental/physical health. The latter includes the use of electroencephalography (EEG) in clinical practice and research. We report a survey of the impact of COVID-19 on the use of clinical EEG in practice and research in several countries, and the recommendations of an international panel of experts for the safe application of EEG during and after this pandemic. METHODS: Fifteen clinicians from 8 different countries and 25 researchers from 13 different countries reported the impact of COVID-19 on their EEG activities, the procedures implemented in response to the COVID-19 pandemic, and precautions planned or already implemented during the reopening of EEG activities. RESULTS: Of the 15 clinical centers responding, 11 reported a total stoppage of all EEG activities, while 4 reduced the number of tests per day. In research settings, all 25 laboratories reported a complete stoppage of activity, with 7 laboratories reopening to some extent since initial closure. In both settings, recommended precautions for restarting or continuing EEG recording included strict hygienic rules, social distance, and assessment for infection symptoms among staff and patients/participants. CONCLUSIONS: The COVID-19 pandemic interfered with the use of EEG recordings in clinical practice and even more in clinical research. We suggest updated best practices to allow safe EEG recordings in both research and clinical settings. The continued use of EEG is important in those with psychiatric diseases, particularly in times of social alarm such as the COVID-19 pandemic.

Clinical utility and applicability of biomarker-based diagnostic criteria for Alzheimer's disease: a BeDeCo survey.

We conducted a survey regarding the medical care of patients with dementia in expert settings in Belgium. Open, unrestricted and motivated answers were centralized, blindly interpreted and structured into categories. The report of the results was then submitted to the participants in subsequent plenary meetings and through email. Fourteen experts responded to the questionnaire, confirming that recent propositions to modify Alzheimer's disease (AD) diagnostic criteria and options have stirred up debate among well-informed and dedicated experts in the field. The opinions were not unanimous and illustrate how difficult it is to find a standardized method of diagnosing this disease. The responses to the survey suggest that application of a step-by-step pragmatic method is used in practice. Only when the combination of clinical findings and classical structural neuro-imaging is insufficient for a diagnosis or suggests an atypical presentation, additional biomarkers are considered. Interestingly, few differences, if any, were observed between the use of biomarkers in MCI and in AD. In conclusion, the Belgian experts consulted in this survey were generally in agreement with the new diagnostic criteria for AD, although some concern was expressed about them being too "amyloidocentric". Although the clinical examination, including a full neuropsychological evaluation, is still considered as the basis for diagnosis, most experts also stated that they use biomarkers to help with diagnosis.

Abnormal proactive and reactive cognitive control during conflict processing in major depression.

According to the Dual Mechanisms of Control framework, cognitive control consists of two complementary components: proactive control refers to anticipatory maintenance of goal-relevant information, whereas reactive control acts as a correction mechanism that is activated when a conflict occurs. Possibly, the well-known diminished inhibitory control in response to negative stimuli in Major Depressive Disorder (MDD) patients stems from a breakdown in proactive control, and/or anomalies in reactive cognitive control. In our study, MDD patients specifically showed increased response latencies when actively inhibiting a dominant response to a sad compared with a happy face. This condition was associated with a longer duration of a dominant ERP topography (800-900 ms poststimulus onset) and a stronger activity in the bilateral dorsal anterior cingulate cortex, reflecting abnormal reactive control when inhibiting attention to a negative stimulus. Moreover, MDD patients showed abnormalities in proactive cognitive control when preparing for the upcoming imperative stimulus (abnormal modulation of the contingent negative variation component), accompanied by more activity in brain regions belonging to the default mode network. All together, deficits to inhibit attention to negative information in MDD might originate from an abnormal use of both proactive resources and reactive control processes.

Electrical brain imaging reveals the expression and timing of altered error monitoring functions in major depression.

Major depressive disorder (MDD) is characterized by disturbances in affect, motivation, and cognitive control processes, including error detection. However, the expression and timing of the impairments during error monitoring remain unclear in MDD. The behavior and event-related brain responses (ERPs) of 20 patients with MDD were compared with those of 20 healthy controls (HCs), while they performed a Go/noGo task. Errors during this task were associated with 2 ERP components, the error-related negativity (ERN/Ne) and the error positivity (Pe). Results show that the ERN/Ne-correct-related negativity (CRN) amplitude difference was significantly larger in MDD patients (after controlling for speed), compared with HCs, although MDD patients exhibited overactive medial frontal cortex (MFC) activation. By comparison, the subsequent Pe component was smaller in MDD patients compared with HCs and this effect was accompanied by a reduced activation of ventral anterior cingulate cortex (ACC) regions. These results suggest that MDD has multiple cascade effects on early error monitoring brain mechanisms.

Multisubject learning for common spatial patterns in motor-imagery BCI.

Motor-imagery-based brain-computer interfaces (BCIs) commonly use the common spatial pattern filter (CSP) as preprocessing step before feature extraction and classification. The CSP method is a supervised algorithm and therefore needs subject-specific training data for calibration, which is very time consuming to collect. In order to reduce the amount of calibration data that is needed for a new subject, one can apply multitask (from now on called multisubject) machine learning techniques to the preprocessing phase. Here, the goal of multisubject learning is to learn a spatial filter for a new subject based on its own data and that of other subjects. This paper outlines the details of the multitask CSP algorithm and shows results on two data sets. In certain subjects a clear improvement can be seen, especially when the number of training trials is relatively low.

On the role of cost-sensitive learning in multi-class brain-computer interfaces.

Brain-computer interfaces (BCIs) present an alternative way of communication for people with severe disabilities. One of the shortcomings in current BCI systems, recently put forward in the fourth BCI competition, is the asynchronous detection of motor imagery versus resting state. We investigated this extension to the three-class case, in which the resting state is considered virtually lying between two motor classes, resulting in a large penalty when one motor task is misclassified into the other motor class. We particularly focus on the behavior of different machine-learning techniques and on the role of multi-class cost-sensitive learning in such a context. To this end, four different kernel methods are empirically compared, namely pairwise multi-class support vector machines (SVMs), two cost-sensitive multi-class SVMs and kernel-based ordinal regression. The experimental results illustrate that ordinal regression performs better than the other three approaches when a cost-sensitive performance measure such as the mean-squared error is considered. By contrast, multi-class cost-sensitive learning enables us to control the number of large errors made between two motor tasks.