Геморрагический инсульт: нейрофизиологические предикторы острого периода

Abstract

Background. Stroke is the most important medical and social problem due to its high proportion of morbidity, disability and mortality among patients of working age.Aims. The aim of the study is to predict the course of the acute period of hemorrhagic parenchymal stroke (supratentorial hemispheric hematomas) based on neuroimaging data (localization, lateralization, hematoma volume) and neurophysiological data.Materials and methods. 86 patients suffering from hemorrhagic stroke of supratentorial localization were examined. The level of consciousness, cognitive functions, and neuroimaging data were evaluated (EEG, heart rate variability, event-related auditory potential). The condition of patients on admission and over time was assessed on the basis of the Glasgow Coma Scale (GCS), the expanded Glasgow Coma Scale and the NIHSS.Results. Based on the results of cluster analysis and expert assessments, two groups of patients were identified: with a relatively favorable and unfavorable course of the acute period of hemorrhagic stroke. Differences in neurophysiological parameters in the groups were established: an increase in the power of theta oscillations and a decrease in the frequency of theta oscillations of the electroencephalogram, a decrease in the amplitude of the N2P2-component of the cognitive evoked P300 potential, an increase in heart rate in an unfavorable course. An artificial neural network has been created to predict the course of the acute period of hemorrhagic stroke upon admission.Conclusion. Machine learning methods allow creating algorithms for predicting the level of consciousness of patients, the acute period of development of intracerebral hematomas of supratentorial localization, the possible development of disease outcomes in patients with non-traumatic intracerebral hematomas based on neurophysiological parameters, as well as the volume of hematoma. A correlate of the unfavorable dynamics turned out to be a reduced bioelectrogenesis in the associative zones of the cortex during stimulus recognition and decision-making, as well as the unfavorable dynamics of the level of consciousness corresponded to a decrease in the amplitude and greater latency of P2N2-peaks, reflecting insufficient activation of the cortical structures during stimulus recognition.

Keywords

• hemorrhagic stroke
• nhiss
• cluster analysis
• neural network