In search of digital approaches and application of artificial intelligence in the study, diagnosis, and treatment of sarcoidosis

Digitalization of healthcare is becoming an integral part of providing medical care to the population. The introduction of artificial intelligence (AI) in medicine leads to the formation of “digital thinking” and public trust in digital healthcare. The purpose of this literature review was to summarize the data related to AI in general and to the study of sarcoidosis – a multiorgan granulomatosis of unknown origin. The most widely represented works are on image recognition, which use different approaches. In pulmonology, this is work with fluorograms, radiographs and computed tomograms. At the same time, comprehensive work is underway on radiomics – comparing image diagnostic data with laboratory and functional data. Programs have been created that recognize speech, analyze the texts of conclusions, the results of tissue diagnostics and even patient auscultation data. In sarcoidosis, the creation of systems to support medical decision-making has been underway since the 1990s, with priority given to Russian phthisiologists, pulmonologists, and mathematicians. In international practice, deep learning has been most fully studied for the diagnosis of pulmonary sarcoidosis. Radiomics was mainly used to differentiate sarcoidosis from malignant tumors. Work is underway to differentiate sarcoidosis and normal data in pulmonary and cardiac sarcoidosis, and for remote self-monitoring of patients. Literature analysis has shown that in clinical medicine, the success of AI is possible only in close cooperation with an expert physician or a multidisciplinary committee of physicians.