Предикторы годовой выживаемости после ишемического инсульта

Неврология, нейропсихиатрия, психосоматика. 2025. Т. 17. № 5. С. 48–54.

Kulikova S., Polyakova I., Kuzmicheva E., Кулеш А. А., Мехряков С. А., Кулеш А. М., Крапивин С. В., Каракулова Ю. В.

Predicting the outcome of ischaemic stroke (IS) is a complex task, as mortality and disability depend on many factors, including age, gender, type and severity of stroke, and comorbidities. Survival rates also vary between countries depending on genetic characteristics and differences in the organisation of healthcare systems.
Objective: to search for predictors of one-year survival after IS in a sample of patients from the Perm region.
Material and methods. The study included 254 patients who had suffered an IS. Seventy-five parameters obtained during routine clinical examination were analysed, including information on the subtype and severity of the stroke, the size and location of the lesion, neurological disorders, comorbidities, and other factors. Relevant features were selected using the WEKA programme, and the selected features were used in a predictive model based on logistic regression.
Results. The following factors have been identified as significant predictors of annual survival in patients who have undergone IS (the sign of the coefficient reflects the relative contribution of the factor to the model and its positive or negative effect): age (-0.02), degree of neurological deficit on the NIHSS scale at discharge (-0.06), haemoglobin level (0.01), infarction in the anterior choroidal artery basin (0.74), recurrent stroke within the following year (-0.02) and cardioembolic stroke subtype (-0.32). The accuracy of the logistic model was 84% with 10-fold cross-validation.
Conclusion. In the model predicting one-year survival after IS, other factors have been identified in addition to age, which is usually associated with a less favourable prognosis. Further multicentre studies are needed to confirm the reliability of the proposed model

Keywords: ischemic stroke, ишемический инсульт, Исход, predictive models, МРТ, предиктивная модель, outcome prediction, Brain MRI