Standardization of implementation of artificial intelligence medical software in terms of digital maturity level

The deployment of artificial intelligence-based (AI-based) software in the clinical practice is a necessary step to improve the quality of public health care. Such implementation requires a comprehensive and balanced approach that, on the one hand, stimulates the introduction of innovations, and, on the other hand, provides reasonable protection for clinicians and patients from potentially untested and low-quality products.The study presents an approach how to choose an optimal scenario for the implementation of AI-based products according to its level of maturity.The products at the initial stage of development, that have not passed independent clinical trials and do not have a registration certificate by Roszdravnadzor, are recommended to choose a pilot project format. It is advisable that sources of financing are based on non-budgetary funds from customers, for example, grants from development institutions or private investments.The products that have confirmed the required level of safety and efficacy and have a registration certificate for a medical device, are advised to launch industrial implementations as decision support systems based on the «second opinion» principle. It is expedient to finance them on the basis of targeted budget programs.The products that have been clinically tested in compliance with the requirements of evidence-based medicine and have an independent assessment of clinical and economic feasibility, may consider wide use in medical diagnostic processes, including the introduction of elements of primary data analysis by autonomous AI agents. It is permissible to employ such products in the procedures and standards for the provision of medical care, as well as clinical recommendations, which will make it possible to pay for the use of such products from compulsory medical insurance funds.