Visual fixation in disorders of consciousness: Development of predictive models to support differential diagnosis

Abstract

The visual fixation represents a doubtful behavioral sign to discriminate Vegetative from Minimally Conscious State (MCS). To disentangle its meaning, we fitted univariate and multivariable logistic regression models matching different neurophysiological and neuroimaging data of 54 patients with Disorders of Consciousness to select the best model predicting which visual performance (visual blink or pursuit) was shown by patients and the best predictors set. The best models found highlighted the importance of the structural MRI and the visual evoked potentials data in predicting visual pursuit. Then, a qualitative pilot test was made on four patients showing visual fixation revealing that the obtained models correctly predict whether the patients' visual performance could support/correlate to a cognitively mediated behavior. The present pilot models could help clinicians to evaluate if the visual fixation response can support the MCS diagnosis.

Keywords: Diagnosis; Minimally conscious state; Predictive modeling; Vegetative state; Visual behavior; Visual system.