“Look at my classifier's result”: Disentangling unresponsive from (minimally) conscious patients

Abstract

Given the fact that clinical bedside examinations can have a high rate of misdiagnosis, machine learning techniques based on neuroimaging and electrophysiological measurements are increasingly being considered for comatose patients and patients with unresponsive wakefulness syndrome, a minimally conscious state or locked-in syndrome. Machine learning techniques have the potential to move from group-level statistical results to personalized predictions in a clinical setting. They have been applied for the purpose of (1) detecting changes in brain activation during functional tasks, equivalent to a behavioral command-following test and (2) estimating signs of consciousness by analyzing measurement data obtained from multiple subjects in resting state. In this review, we provide a comprehensive overview of the literature on both approaches and discuss the translation of present findings to clinical practice. We found that most studies struggle with the difficulty of establishing a reliable behavioral assessment and fluctuations in the patient's levels of arousal. Both these factors affect the training and validation of machine learning methods to a considerable degree. In studies involving more than 50 patients, small to moderate evidence was found for the presence of signs of consciousness or good outcome, where one study even showed strong evidence for good outcome.

Degrees of functional connectome abnormality in disorders of consciousness

“…Several other recent studies attempted to predict the MCS and UWS diagnosis based on different types of biomarkers from PET, DTI, EEG, and fMRI data and showed promising results, reviewed in Noirhomme, Brecheisen, Lesenfants, Antonopoulos, and Laureys (). Due to limited sample sizes, the confidence intervals for sensitivity and specificity are wide and overlap considerably.…”

Tracking attention in a visual active paradigm for the diagnosis of disorders of consciousness

“…Visual (active red versus active yellow) and attentional (active versus passive) classification performances were then computed with a linear discriminant analysis (LDA), and assessed with a 10x10-fold cross-validation. A permutation test (Nichols & Holmes 2002;Noirhomme et al 2017) evaluated the chance-level for each participant (1000 repetitions, LDA classification, 10x10-fold cross validation, p<0.01). The significance of change between conditions was assessed with a non-parametric Wilcoxon signed-rank test (2-tailed, p < 0.01), with Bonferroni correction for multiple comparisons.…”

The Clinical Diagnostic Utility of Electrophysiological Techniques in Assessment of Patients With Disorders of Consciousness Following Acquired Brain Injury: A Systematic Review

“…36,37 Such electrophysiological features or activation patterns can also be applied in machine learning systems that allow quantification of differences in neural responses at an individual level. 38,39 Surface electromyogram (EMG) is, on the other hand, recordings of electrical activity in muscles, and is a commonly used tool to study physiological principles of muscles related to movement generation. 40,41…”

“…44,51 Yet other papers only provide a topical overview without explicit systematic literature search strategies. 38,46,48,49,[52][53][54] In addition, no existing review provides an overview over the rate of excluded subjects across studies due to methodological artifacts, which is quite common in electrophysiological methods in general, and might be expected to be even higher in groups known to have ample muscle artifact, and lack cooperative abilities in the engaged test-situation.…”

“…Notably, the number of patients showing electrophysiological signs of mental effort despite lack of behavioral command-following, is in line with those obtained in fMRI studies using active tasks. 38,94 In summary, the two large studies applying multivariate EEG-classifier systems most likely represent the method with best balance between rates of sensitivity and specificity.…”