14 | EEG markers of motion-induced kinetosis: quantifying spectral changes and individual susceptibility

Schulz HB, Schwab M, Müller S, Schönfeld U, Zacharias N | ENT Department - CBF, Charité - Universitätsmedizin Berlin & Faculty of Mechanical Engineering and Transport, Technical University Berlin, Germany

Background: Conflict theories are widely accepted as a key mechanism underlying motion sickness (kinetosis), but objective physiological studies on cortical activity during motion-induced kinetosis remain limited. This study investigates the neuronal correlates associated with the development of motion-induced kinetosis, provoked by a rotating chair paradigm. The aim was to identify specific spectral electroencephalographic (EEG) changes related to symptom intensities.

Materials and Methods: Participants (N=19) underwent the Staircase Velocity Motion Test (Calkins et al.,1987) to induce kinetosis by provoking vestibular conflict. Subjective symptom intensity was assessed every minute by the Misery Scale index, MISC (Bos et al.,2005). Spectral analysis focused on power changes across standard frequency bands assessed using 32-channel EEG. A machine learning approach (XGBoost) using EEG features to predict kinetosis severity was performed.

Results: Analysis revealed a significant increase in alpha band power that correlated with increasing MISC scores, starting in posterior parietal cortex for lower (0-3) and evolving to fronto-central cortex for higher MISC scores (4-7). Transient increase of delta and theta power of vestibular-related cortical regions were prominent for lower MISC scores (1-2). Prior to the onset of nausea (MISC 5) delta and theta power peaked. At MISC 6, delta and theta power of vestibular cortical regions as well as parieto-temporal gamma power significantly decreased.

Conclusions: Our data may suggest three consecutive physiological states of kinetosis corresponding to MISC ranges 0-2, 3-5, and 6-8, although further analysis is needed. Machine learning results suggest potential of an EEG-based prediction model as preliminary data show an accuracy of 0.84.