Artificial intelligence on the doorstep of the laboratory

ABSRACT

The review focuses on artificial intelligence (AI), a set of technological solutions with an emphasis on machine learning (ML). Aspects of laboratory performance with typical AI-based solutions are listed. Examples of research using AI technologies accelerating the development of personalized medicine, the search for new biomarkers and scientific information are given. The applications of AI are considered: 1) in demand management, when a model based on ML recommends a list of laboratory tests, predicts the category of test values from the results of other tests; 2) in the analytical stage, as an aid in quality control and replacement of conventional microscopy with digital image analysis; 3) in interpretation of lab results using medical decision support systems. Examples of application of ML, indicating its high accuracy and speed compared with traditional methods, for diagnosis, risk assessment, prognosis in oncology, cardiology, hematology, in sepsis, neurodegenerative and autoimmune diseases are presented. The adoption of AI technologies in medicine is accelerating, raising issues of ethics, regulation, data quality, competence and willingness to change for medical personnel. As of this writing, more than 470 AI-based medical devices have been registered in the United States and Europe, and 12 software medical devices have been registered in Russia. This requires the responsibility of laboratory specialists in providing standardized and quality data, proactive problem-setting, cooperation with IT specialists and doctors of clinical specialties.