Аннотация:For a set of circulatory pathologies, virtual databases for learning and diagnosticsare formed by using mathematical models of blood flow by approximation ofquasione-dimensional hemodynamics. The procedure is applied to diagnosing blood flow pathologies based on thepatient’s pulse wave data and permits prior computer modelling of AI methods now inuse for that purpose. Feasibility and practicality are demonstrated. As implementation and verification of AI methods require significant time,involve high material costs, and successful results are still not guaranteed, the approachis suitable for prior evaluation of the accuracy to be expected from AI in this context.