RESUMO
The aim of the study is to assess the possibility of using artificial intelligence to determine the most significant predictors of the operative correction outcomes for patients with damaged coronary and carotid arteries. Materials and Methods: The retrospective study of the simultaneous (or single-stage) surgical intervention results has been carried out in patients with combined atherosclerotic damage of the coronary bed and cerebral arteries (n=42), which was severe and extensive. The parameters which may be predictors of the cardiovascular risk were analyzed using the TADA program. Ten models were built for program learning. The model with 92% predictive accuracy appeared to be the most successful. Results: Simultaneous correction resulted in the absence of 30-day coronary complications in all patients. With respect to the cerebral vascular territory, acute ischemic stroke developed in 2 patients. The lethality rate was 2.4%, the fatal outcome was caused by postoperative gastrointestinal bleeding.The TADA program model considered the following parameters to be the most significant predictors: internal carotid artery cross-clamping time in minutes (51.24%); damage to the left coronary artery stem (30.42%); diastolic AP (18.28%). If cross-clamping of the internal carotid artery lasts for less than 18 min, complications are not likely to occur, while they are practically inevitable if the time exceeds 46 min. The probability of complications grows nonlinearly with the increase of the extent of the left coronary artery stem injury. A high diastolic AP never virtually coincides with the presence of complications, nor does the low one. The highest probability of complications is at the values from 70 to 80 mm Hg.In patients with a triple vessel injury of the coronary arteries, a representative picture of a nonsignificant feature is observed. Conclusion: Application of artificial intelligence for determining risk predictors for patients with concurrent atherosclerotic damage of the coronary and carotid arteries is an effective method for prognosticating the risks of simultaneous interventions.
