The purpose of the study was to investigate the possibility of applying first-reader and second-reader modes in the implementation of an automatic detection program for MCA ischemic stroke in the diagnostic process of radiologists with less than 3 years of experience and varying expertise in emergency neuroradiology. Material and methods. The study included a software product based on artificial intelligence technologies, as well as seven doctors with less than 3 years of experience and varying expertise in the diagnosis of ischemic stroke. Complementary evaluation was performed based on a cohort of 100 patients admitted to the regional vascular center in Saint Petersburg with clinical presentation of ischemic stroke in the territory of the middle cerebral artery, who underwent native CT brain studies. Ischemic stroke was confirmed in half of the patients based on clinical data, as well as CT angiography of the cerebral vessels and CT perfusion. The diagnosis was ruled out in the other half. Two variants of implementing the artificial intelligence algorithm as a decision support system in the diagnostic process of a radiologist were simulated: the first (parallel) and second reader modes. Results. The results of the study showed that the application of the complementary evaluation parallel- reader mode leads to an increase in diagnostic efficiency indicators and interobserver agreement in assessing ASPECTS scale among young specialists, regardless of their experience with urgent pathology.