The effectiveness of using artificial intelligence in clinical medicine

Abstract

Objective : To examine the effectiveness of using artificial intelligence (AI) in clinical medicine (in terms of accuracy, sensitivity, and specificity). Material and methods . The study involved a search and analysis of scientific publications examining various types of AI training and training approaches, as well as areas of AI application in clinical practice, which were submitted to PubMed/MEDLINE, Scopus, Web of Science, Embase, eLibrary, and CyberLeninka databases in 2009–2023. A sequential analysis of randomly sampled articles yielded 30 publications on the use of AI in endocrinology (4 articles), dermatovenerology (3), cardiology (1), radiology (1), gastroenterology (1), neurology (5), hematology (5), nephrology (4), orthopedics and rheumatology (4), oncology (2). Results . AI demonstrated sufficient effectiveness: accuracy ranged from 49% to 99%, sensitivity from 42% to 100%, and specificity from 48% to 100% in such areas as cardiology, endocrinology, gastroenterology, dermatovenereology, and radiology. In some cases, AI was more effective than clinical diagnostics by medical specialists, e.g. in detecting melanoma and diagnosing atrial fibrillation. Conclusion . AI shows high diagnostic efficiency, increases accuracy and speeds up diagnostic search, which makes it promising to expand the use of AI in clinical medicine.

Ähnliche Arbeiten