Critical evaluation of auditory event-related potential deficits in schizophrenia: evidence from large-scale single-subject pattern classification

Does temporal irregularity drive prediction failure in schizophrenia? temporal modelling of ERPs

Schizophrenia subjects often suffer from a failure to properly predict incoming inputs; most notably, some patients exhibit impaired prediction of the sensory consequences of their own actions. The mechanisms underlying this deficit remain unclear, though. One possible mechanism could consist in aberrant predictive processing, as schizophrenic patients show relatively less attenuated neuronal activity to self-produced tones, than healthy controls. Here, we tested the hypothesis that this aberrant predictive mechanism would manifest itself in the temporal irregularity of neuronal signals. For that purpose, we here introduce an event-related potential (ERP) study model analysis that consists of an EEG real-time model equation, eeg(t) and a frequency Laplace transformed Transfer Function (TF) equation, eeg(s). Combining circuit analysis with control and cable theory, we estimate the temporal model representations of auditory ERPs to reveal neural mechanisms that make predictions about self-generated sensations. We use data from 49 schizophrenic patients (SZ) and 32 healthy control (HC) subjects in an auditory ‘prediction’ paradigm; i.e., who either pressed a button to deliver a sound tone (epoch a), or just heard the tone without button press (epoch b). Our results show significantly higher degrees of temporal irregularity or imprecision between different trials of the ERP from the Cz electrode (N100, P200) in SZ compared to HC (Levene’s test, p < 0.0001) as indexed by altered latency, lower similarity/correlation of single trial time courses (using dynamic time warping), and longer settling times to reach steady state in the intertrial interval. Using machine learning, SZ vs HC could be highly accurately classified (92%) based on the temporal parameters of their ERPs’ TF models, using as features the poles of the TF rational functions. Together, our findings show temporal irregularity or imprecision between single trials to be abnormally increased in SZ. This may indicate a general impairment of SZ, related to precisely predicting the sensory consequences of one’s actions.

Unveiling the Associations between EEG Indices and Cognitive Deficits in Schizophrenia-Spectrum Disorders: A Systematic Review

Cognitive dysfunctions represent a core feature of schizophrenia-spectrum disorders due to their presence throughout different illness stages and their impact on functioning. Abnormalities in electrophysiology (EEG) measures are highly related to these impairments, but the use of EEG indices in clinical practice is still limited. A systematic review of articles using Pubmed, Scopus and PsychINFO was undertaken in November 2021 to provide an overview of the relationships between EEG indices and cognitive impairment in schizophrenia-spectrum disorders. Out of 2433 screened records, 135 studies were included in a qualitative review. Although the results were heterogeneous, some significant correlations were identified. In particular, abnormalities in alpha, theta and gamma activity, as well as in MMN and P300, were associated with impairments in cognitive domains such as attention, working memory, visual and verbal learning and executive functioning during at-risk mental states, early and chronic stages of schizophrenia-spectrum disorders. The review suggests that machine learning approaches together with a careful selection of validated EEG and cognitive indices and characterization of clinical phenotypes might contribute to increase the use of EEG-based measures in clinical settings.

Machine Learning Techniques for the Diagnosis of Schizophrenia Based on Event-Related Potentials

Antecedent

The event-related potential (ERP) components P300 and mismatch negativity (MMN) have been linked to cognitive deficits in patients with schizophrenia. The diagnosis of schizophrenia could be improved by applying machine learning procedures to these objective neurophysiological biomarkers. Several studies have attempted to achieve this goal, but no study has examined Multiple Kernel Learning (MKL) classifiers. This algorithm finds optimally a combination of kernel functions, integrating them in a meaningful manner, and thus could improve diagnosis.

Objective

This study aimed to examine the efficacy of the MKL classifier and the Boruta feature selection method for schizophrenia patients (SZ) and healthy controls (HC) single-subject classification.

Methods

A cohort of 54 SZ and 54 HC participants were studied. Three sets of features related to ERP signals were calculated as follows: peak related features, peak to peak related features, and signal related features. The Boruta algorithm was used to evaluate the impact of feature selection on classification performance. An MKL algorithm was applied to address schizophrenia detection.

Results

A classification accuracy of 83% using the whole dataset, and 86% after applying Boruta feature selection was obtained. The variables that contributed most to the classification were mainly related to the latency and amplitude of the auditory P300 paradigm.

Conclusion

This study showed that MKL can be useful in distinguishing between schizophrenic patients and controls when using ERP measures. Moreover, the use of the Boruta algorithm provides an improvement in classification accuracy and computational cost.

Importance of the Features of Event-Related Potentials Used for a Machine Learning-Based Model Applied to Single-Trial Data during Oddball Task

In this study, a method for assessing the human state and brain-machine interface (BMI) has been developed using event-related potentials (ERPs). Most of these algorithms are classified based on the ERP characteristics. To observe the characteristics of ERPs, an averaging method using electroencephalography (EEG) signals cut out by time-locking to the event for each condition is required. To date, several classification methods using only single-trial EEG signals have been studied. In some cases, the machine learning models were used for the classifications; however, the relationship between the constructed model and the characteristics of ERPs remains unclear. In this study, the LightGBM model was constructed for each individual to classify a single-trial waveform and visualize the relationship between these features and the characteristics of ERPs. The features used in the model were the average values and standard deviation of the EEG amplitude with a time width of 10 ms. The best area under the curve (AUC) score was 0.92, but, in some cases, the AUC scores were low. Large individual differences in AUC scores were observed. In each case, on checking the importance of the features, high importance was shown at the 10-ms time width section, where a large difference was observed in ERP waveforms between the target and the non-target. Since the model constructed in this study was found to reflect the characteristics of ERP, as the next step, we would like to try to improve the discrimination performance by using stimuli that the participants can concentrate on with interest.

Schizophrenia: A Survey of Artificial Intelligence Techniques Applied to Detection and Classification

Artificial Intelligence in healthcare employs machine learning algorithms to emulate human cognition in the analysis of complicated or large sets of data. Specifically, artificial intelligence taps on the ability of computer algorithms and software with allowable thresholds to make deterministic approximate conclusions. In comparison to traditional technologies in healthcare, artificial intelligence enhances the process of data analysis without the need for human input, producing nearly equally reliable, well defined output. Schizophrenia is a chronic mental health condition that affects millions worldwide, with impairment in thinking and behaviour that may be significantly disabling to daily living. Multiple artificial intelligence and machine learning algorithms have been utilized to analyze the different components of schizophrenia, such as in prediction of disease, and assessment of current prevention methods. These are carried out in hope of assisting with diagnosis and provision of viable options for individuals affected. In this paper, we review the progress of the use of artificial intelligence in schizophrenia.

Advanced EEG-based learning approaches to predict schizophrenia: Promises and pitfalls

The complexity and heterogeneity of schizophrenia symptoms challenge an objective diagnosis, which is typically based on behavioral and clinical manifestations. Moreover, the boundaries of schizophrenia are not precisely demarcated from other nosologic categories, such as bipolar disorder. The early detection of schizophrenia can lead to a more effective treatment, improving patients' quality of life. Over the last decades, hundreds of studies aimed at specifying the neurobiological mechanisms that underpin clinical manifestations of schizophrenia, using techniques such as electroencephalography (EEG). Changes in event-related potentials of the EEG have been associated with sensory and cognitive deficits and proposed as biomarkers of schizophrenia. Besides contributing to a more effective diagnosis, biomarkers can be crucial to schizophrenia onset prediction and prognosis. However, any proposed biomarker requires substantial clinical research to prove its validity and cost-effectiveness. Fueled by developments in computational neuroscience, automatic classification of schizophrenia at different stages (prodromal, first episode, chronic) has been attempted, using brain imaging pattern recognition methods to capture differences in functional brain activity. Advanced learning techniques have been studied for this purpose, with promising results. This review provides an overview of recent machine learning-based methods for schizophrenia classification using EEG data, discussing their potentialities and limitations. This review is intended to serve as a starting point for future developments of effective EEG-based models that might predict the onset of schizophrenia, identify subjects at high-risk of psychosis conversion or differentiate schizophrenia from other disorders, promoting more effective early interventions.

Meta-analysis of auditory P50 sensory gating in schizophrenia and bipolar disorder

The ability of the brain to reduce the amount of trivial or redundant sensory inputs is called gating function. Dysfunction of sensory gating may lead to cognitive fragmentation and poor real-world functioning. The auditory dual-click paradigm is a pertinent neurophysiological measure of sensory gating function. This meta-analysis aimed to examine the subcomponents of abnormal P50 waveforms in bipolar disorder and schizophrenia to assess P50 sensory gating deficits and examine effects of diagnoses, illness states (first-episode psychosis vs. schizophrenia, remission vs. episodes in bipolar disorder), and treatment status (medication-free vs. medicated). Literature search of PubMed between Jan 1st 1980 and March 31st 2019 identified 2091 records for schizophrenia, 362 for bipolar disorder. 115 studies in schizophrenia (4932 patients), 16 in bipolar disorder (975 patients) and 10 in first-degree relatives (848 subjects) met the inclusion criteria. P50 sensory gating ratio (S2/S1) and S1-S2 difference were significantly altered in schizophrenia, bipolar disorder and their first-degree relatives. First-episode psychosis did not differ from schizophrenia, however episodes altered P50 sensory gating in bipolar disorder. Medications improve P50 sensory gating alterations in schizophrenia significantly and at trend level in bipolar disorder. Future studies should examine longitudinal course of P50 sensory gating in schizophrenia and bipolar disorder.

Feature optimization method for machine learning-based diagnosis of schizophrenia using magnetoencephalography

BACKGROUND

When many features and a small number of clinical data exist, previous studies have used a few top-ranked features from the Fisher's discriminant ratio (FDR) for feature selection. However, there are many similarities between selected features. New method: To reduce the redundant features, we applied a technique employing FDR in conjunction with feature correlation. We performed an attention network test on schizophrenic patients and normal subjects with a 152-channel magnetoencephalograph. P300m amplitudes of event-related fields (ERFs) were used as features at the sensor level and P300m amplitudes of ERFs for 500 nodes on the cortex surface were used as features at the source level. Features were ranked using FDR criterion and cross-correlation measure, and then the highest ranked 10 features were selected and an exhaustive search was used to find combination having the maximum accuracy.

RESULTS

At the sensor level, we found a single channel of the occipital region that distinguished the two groups with an accuracy of 89.7 %. At source level, we obtained an accuracy of 96.2 % using two features, the left superior frontal region and the left inferior temporal region.

COMPARISON WITH EXISTING METHOD

At source level, we obtained a higher accuracy than traditional method using only FDR criterion (accuracy = 88.5 %). We used only the P300 m amplitude (not latency) on a single channel and two brain regions at a fairly high rate.

P50, N100, and P200 Auditory Sensory Gating Deficits in Schizophrenia Patients

BACKGROUND

Sensory gating describes neurological processes of filtering out redundant or unnecessary stimuli during information processing, and sensory gating deficits may contribute to the symptoms of schizophrenia. Among the three components of auditory event-related potentials reflecting sensory gating, P50 implies pre-attentional filtering of sensory information and N100/P200 reflects attention triggering and allocation processes. Although diminished P50 gating has been extensively documented in patients with schizophrenia, previous studies on N100 were inconclusive, and P200 has been rarely examined. This study aimed to investigate whether patients with schizophrenia have P50, N100, and P200 gating deficits compared with control subjects.

METHODS

Control subjects and clinically stable schizophrenia patients were recruited. The mid-latency auditory evoked responses, comprising P50, N100, and P200, were measured using the auditory-paired click paradigm without manipulation of attention. Sensory gating parameters included S1 amplitude, S2 amplitude, amplitude difference (S1-S2), and gating ratio (S2/S1). We also evaluated schizophrenia patients with PANSS to be correlated with sensory gating indices.

RESULTS

One hundred four patients and 102 control subjects were examined. Compared to the control group, schizophrenia patients had significant sensory gating deficits in P50, N100, and P200, reflected by larger gating ratios and smaller amplitude differences. Further analysis revealed that the S2 amplitude of P50 was larger, while the S1 amplitude of N100/P200 was smaller, in schizophrenia patients than in the controls. We found no correlations between sensory gating indices and schizophrenia positive or negative symptom clusters. However, we found a negative correlation between the P200 S2 amplitude and Bell's emotional discomfort factor/Wallwork's depressed factor.

CONCLUSION

Till date, this study has the largest sample size to analyze P50, N100, and P200 collectively by adopting the passive auditory paired-click paradigm without distractors. With covariates controlled for possible confounds, such as age, education, smoking amount and retained pairs, we found that schizophrenia patients had significant sensory gating deficits in P50-N100-P200. The schizophrenia patients had demonstrated a unique pattern of sensory gating deficits, including repetition suppression deficits in P50 and stimulus registration deficits in N100/200. These results suggest that sensory gating is a pervasive cognitive abnormality in schizophrenia patients that is not limited to the pre-attentive phase of information processing. Since P200 exhibited a large effect size and did not require additional time during recruitment, future studies of P50-N100-P200 collectively are highly recommended.

Altered individual behavioral and EEG parameters are related to the EDSS score in relapsing-remitting multiple sclerosis patients

Functional neuroanatomy of cognitive impairment in multiple sclerosis is currently still a challenge. During the progression of the disease, several cognitive mechanisms deteriorate thus diminishing the patient’s quality of life. A primary objective in the cognitive assessment of multiple sclerosis (MS) patients is to find reliable measures utilizing diverse neuroimaging techniques. Moreover, especially relevant in the clinical environment is finding technical approaches that could be applied to individual participants and not only for group analysis. A 64-channel electroencephalographic recording (EEG) was made with thirty participants divided into three groups of equivalent size (N = 10) (healthy control, low-EDSS (1–2.5) and moderate-EDSS (4–6)). Correlation analysis was applied to multiple measures: behavior, neuropsychological tests (Paced Auditory Serial Addition Test, 3 seconds (PASAT-3s) and the Symbol Digit Modality Test (SDMT)), Expanded Disability Status Scale (EDSS), even-related potential (P3) and event-related desynchronization (ERD) parameters and the correlation scores between individual participant’s P3/ERD maps and the healthy grand average P3/ERDmaps. Statistical analysis showed that diverse parameters exhibited significant correlations. A remarkable correlation was the moderate score found between SDMT and EDSS (r = −0.679, p = 0.0009). However, the strongest correlation was between the value of integrated measures (reaction time, P3 and ERD latency) and EDSS (r = 0.699, p = 0.0006). In regard to correlations for grand average maps between groups, the P3 component exhibited a lower score according to a more deteriorated condition (higher EDSS). In contrast, ERD maps remained stable with an increase of EDSS. Lastly, a Z-transformation of individual values of all variables included in the study exhibited heterogeneity in cognitive alterations in the multiple sclerosis participants.

The coupling of low-level auditory dysfunction and oxidative stress in psychosis patients

Patients diagnosed with schizophrenia often present with low-level sensory deficits. It is an open question whether there is a functional link between these deficits and the pathophysiology of the disease, e.g. oxidative stress and glutathione (GSH) metabolism dysregulation. Auditory evoked potentials (AEPs) were recorded from 21 psychosis disorder patients and 30 healthy controls performing an active, auditory oddball task. AEPs to standard sounds were analyzed within an electrical neuroimaging framework. A peripheral measure of participants' redox balance, the ratio of glutathione peroxidase and glutathione reductase activities (GPx/GR), was correlated with the AEP data. Patients displayed significantly decreased AEPs over the time window of the P50/N100 complex resulting from significantly weaker responses in the left temporo-parietal lobe. The GPx/GR ratio significantly correlated with patients' brain activity during the time window of the P50/N100 in the medial frontal lobe. We show for the first time a direct coupling between electrophysiological indices of AEPs and peripheral redox dysregulation in psychosis patients. This coupling is limited to stages of auditory processing that are impaired relative to healthy controls and suggests a link between biochemical and sensory dysfunction. The data highlight the potential of low-level sensory processing as a trait-marker of psychosis.

A Computer-Aided Diagnosis System With EEG Based on the P3b Wave During an Auditory Odd-Ball Task in Schizophrenia

OBJECTIVE

To design a Computer-aided diagnosis (CAD) system using an optimized methodology over the P3b wave in order to objectively and accurately discriminate between healthy controls (HC) and schizophrenic subjects (SZ).

METHODS

We train, test, analyze, and compare various machine learning classification approaches optimized in terms of the correct classification rate (CCR), the degenerated Youden's index (DYI) and the area under the receiver operating curve (AUC). CAD system comprises five stages: electroencephalography (EEG) preprocessing, feature extraction, seven electrode groupings, discriminant feature selection, and binary classification.

RESULTS

With two optimal combinations of electrode grouping, filtering, feature selection algorithm, and classification machine, we get either a mean CCR = 93.42%, specificity = 0.9673, sensitivity = 0.8727, DYI = 0.9188, and AUC = 0.9567 (total-15 Hz-J5-MLP), or a mean CCR = 92.23%, specificity = 0.9499, sensitivity = 0.8838, DYI = 0.9162, and AUC = 0.9807 (right hemisphere-35 Hz-J5-SVM), which to our knowledge are higher than those available to date.

CONCLUSIONS

We have verified that a more restrictive low-pass filtering achieves higher CCR as compared to others at higher frequencies in the P3b wave. In addition, results validate previous hypothesis about the importance of the parietal-temporal region, associated with memory processing, allowing us to identify powerful {feature,electrode} pairs in the diagnosis of schizophrenia, achieving higher CCR and AUC in classification of both right and left Hemispheres, and parietal-temporal EEG signals, like, for instance, the {PSE, P4} pair (J5 and mutual information feature selection).

SIGNIFICANCE

Diagnosis of schizophrenia is made thoroughly by psychiatrists but as any human-based decision that has a subjective component. This CAD system provides the human expert with an objective complimentary measure to help him in diagnosing schizophrenia.

Auditory prediction errors as individual biomarkers of schizophrenia

Schizophrenia is a complex psychiatric disorder, typically diagnosed through symptomatic evidence collected through patient interview. We aim to develop an objective biologically-based computational tool which aids diagnosis and relies on accessible imaging technologies such as electroencephalography (EEG). To achieve this, we used machine learning techniques and a combination of paradigms designed to elicit prediction errors or Mismatch Negativity (MMN) responses. MMN, an EEG component elicited by unpredictable changes in sequences of auditory stimuli, has previously been shown to be reduced in people with schizophrenia and this is arguably one of the most reproducible neurophysiological markers of schizophrenia.

EEG data were acquired from 21 patients with schizophrenia and 22 healthy controls whilst they listened to three auditory oddball paradigms comprising sequences of tones which deviated in 10% of trials from regularly occurring standard tones. Deviant tones shared the same properties as standard tones, except for one physical aspect: 1) duration - the deviant stimulus was twice the duration of the standard; 2) monaural gap - deviants had a silent interval omitted from the standard, or 3) inter-aural timing difference, which caused the deviant location to be perceived as 90° away from the standards.

We used multivariate pattern analysis, a machine learning technique implemented in the Pattern Recognition for Neuroimaging Toolbox (PRoNTo) to classify images generated through statistical parametric mapping (SPM) of spatiotemporal EEG data, i.e. event-related potentials measured on the two-dimensional surface of the scalp over time. Using support vector machine (SVM) and Gaussian processes classifiers (GPC), we were able classify individual patients and controls with balanced accuracies of up to 80.48% (p-values = 0.0326, FDR corrected) and an ROC analysis yielding an AUC of 0.87. Crucially, a GP regression revealed that MMN predicted global assessment of functioning (GAF) scores (correlation = 0.73, R² = 0.53, p = 0.0006).

•

The diagnostic utility of multiple auditory oddball stimulus paradigms is assessed.

•

Greatest classification accuracy was achieved using a monaural gap stimulus paradigm.

•

The full post-stimulus epoch contains relevant discriminatory components.

N100 Repetition Suppression Indexes Neuroplastic Defects in Clinical High Risk and Psychotic Youth

Highly penetrant mutations leading to schizophrenia are enriched for genes coding for N-methyl-D-aspartate receptor signaling complex (NMDAR-SC), implicating plasticity defects in the disease's pathogenesis. The importance of plasticity in neurodevelopment implies a role for therapies that target these mechanisms in early life to prevent schizophrenia. Testing such therapies requires noninvasive methods that can assess engagement of target mechanisms. The auditory N100 is an obligatory cortical response whose amplitude decreases with tone repetition. This adaptation may index the health of plasticity mechanisms required for normal development. We exposed participants aged 5 to 17 years with psychosis (n = 22), at clinical high risk (CHR) for psychosis (n = 29), and healthy controls (n = 17) to an auditory tone repeated 450 times and measured N100 adaptation (mean amplitude during first 150 tones − mean amplitude during last 150 tones). N100 adaptation was reduced in CHR and psychosis, particularly among participants <13 years old. Initial N100 blunting partially accounted for differences. Decreased change in the N100 amplitude with tone repetition may be a useful marker of defects in neuroplastic mechanisms measurable early in life.

Single-subject prediction of response inhibition behavior by event-related potentials

Much research has been devoted to investigating response inhibition and the neuronal processes constituting this essential cognitive faculty. However, the nexus between cognitive subprocesses, behavior, and electrophysiological processes remains associative in nature. We therefore investigated whether neurophysiological correlates of inhibition subprocesses merely correlate with behavioral performance or actually provide information expedient to the prediction of behavior on a single-subject level. Tackling this question, we used different data-driven classification approaches in a sample of n = 262 healthy young subjects who completed a standard Go/Nogo task while an EEG was recorded. On the basis of median-split response inhibition performance, subjects were classified as "accurate/slow" and "less accurate/fast." Even though these behavioral group differences were associated with significant amplitude variations in classical electrophysiological correlates of response inhibition (i.e., N2 and P3), they were not predictive for group membership on a single-subject level. Instead, amplitude differences in the Go-P2 originating in the precuneus (BA7) were shown to predict group membership on a single-subject level with up to 64% accuracy. These findings strongly suggest that the behavioral outcome of response inhibition greatly depends on the amount of cognitive resources allocated to early stages of stimulus-response activation during responding. This suggests that research should focus more on early processing steps during responding when trying to understand the origin of interindividual differences in response inhibition processes.

Early auditory processing evoked potentials (N100) show a continuum of blunting from clinical high risk to psychosis in a pediatric sample

BACKGROUND

The N100 is a negative deflection in the surface EEG approximately 100 ms after an auditory signal. It has been shown to be reduced in individuals with schizophrenia and those at clinical high risk (CHR). N100 blunting may index neural network dysfunction underlying psychotic symptoms. This phenomenon has received little attention in pediatric populations.

METHOD

This cross-sectional study compared the N100 response measured via the average EEG response at the left medial frontal position FC1 to 150 sinusoidal tones in participants ages 5 to 17 years with a CHR syndrome (n=29), a psychotic disorder (n=22), or healthy controls (n=17).

RESULTS

Linear regression analyses that considered potential covariates (age, gender, handedness, family mental health history, medication usage) revealed decreasing N100 amplitude with increasing severity of psychotic symptomatology from healthy to CHR to psychotic level.

CONCLUSIONS

Longitudinal assessment of the N100 in CHR children who do and do not develop psychosis will inform whether it predicts transition to psychosis and if its response to treatment predicts symptom change.

Schizophrenia detection and classification by advanced analysis of EEG recordings using a single electrode approach

Electroencephalographic (EEG) analysis has emerged as a powerful tool for brain state interpretation and diagnosis, but not for the diagnosis of mental disorders; this may be explained by its low spatial resolution or depth sensitivity. This paper concerns the diagnosis of schizophrenia using EEG, which currently suffers from several cardinal problems: it heavily depends on assumptions, conditions and prior knowledge regarding the patient. Additionally, the diagnostic experiments take hours, and the accuracy of the analysis is low or unreliable. This article presents the "TFFO" (Time-Frequency transformation followed by Feature-Optimization), a novel approach for schizophrenia detection showing great success in classification accuracy with no false positives. The methodology is designed for single electrode recording, and it attempts to make the data acquisition process feasible and quick for most patients.

Single-subject classification of schizophrenia patients based on a combination of oddball and mismatch evoked potential paradigms

OBJECTIVE

The diagnostic process for schizophrenia is mainly clinical and has to be performed by an experienced psychiatrist, relying primarily on clinical signs and symptoms. Current neurophysiological measurements can distinguish groups of healthy controls and groups of schizophrenia patients. Individual classification based on neurophysiological measurements mostly shows moderate accuracy. We wanted to examine whether it is possible to distinguish controls and patients individually with a good accuracy. To this end we used a combination of features extracted from the auditory and visual P300 paradigms and the mismatch negativity paradigm.

METHODS

We selected 54 patients and 54 controls, matched for age and gender, from the data available at the UPC Kortenberg. The EEG-data were high- and low-pass filtered, epoched and averaged. Features (latencies and amplitudes of component peaks) were extracted from the averaged signals. The resulting dataset was used to train and test classification algorithms. First on separate paradigms and then on all combinations, we applied Naïve Bayes, Support Vector Machine and Decision Tree, with two of its improvements: Adaboost and Random Forest.

RESULTS

For at least two classifiers the performance increased significantly by combining paradigms compared to single paradigms. The classification accuracy increased from at best 79.8% when trained on features from single paradigms, to 84.7% when trained on features from all three paradigms.

CONCLUSION

A combination of features originating from three evoked potential paradigms allowed us to accurately classify individual subjects as either control or patient. Classification accuracy was mostly above 80% for the machine learners evaluated in this study and close to 85% at best.