Theoretical issues of legal regulation and prospects for the use of artificial intelligence in Ukraine in conducting medical research

Abstract

Artificial intelligence technologies are present in both everyday and professional life, are used in various areas, but at the moment there is no clear legal definition and legal regulation in the current legislation of Ukraine, responsibility for the use of artificial intelligence is not regulated. Currently, in Ukraine, the regulation of a computer program is equated with a literary work, therefore, legal regulation of artificial intelligence is recommended from the point of view of the norms of civil law of Ukraine. There is no clear path for the implementation and integration of generative artificial intelligence in the provision of medical services. The purpose of the study was to compare foreign and domestic principles of legal regulation of the use of artificial intelligence, to study the effectiveness of the use and prospects for further implementation of artificial intelligence in medical (pathohistological) research by analyzing modern scientific publications. The search was conducted in the Google search engine (https://www.google.com.ua/), the US National Library of Medicine database (https://www.ncbi.nlm.nih.gov/), the PubMed database (https://pubmed.ncbi.nlm.nih.gov), the ScienceDirect medical literature platform (https://sciencedirect.com/) and Elsevier (https://www.elsevier.com/). A total of 115 publications were reviewed, of which 69 were selected, relating to the historical and legal aspects of the use of artificial intelligence, research on the results of the implementation of digital pathology, artificial intelligence in pathology and cytomorphology. Pathology relies on the recognition of microscopic morphological features, image analysis plays a crucial role, allowing the identification, categorization and characterization of tissue types, cells and their changes. Traditional methods of pathology are time-consuming and sensitive to inter-observer variability. Digitization of pathohistological images through objective and effective computer analysis methods has provided additional tools for faster, better and more accurate diagnosis. The effective use of artificial intelligence in the diagnosis of individual neoplasms, assessment of the molecular subtype of breast cancer, urine cytology according to the Paris system, determination of Gleason scores for prostate cancer, assessment of PD-L1 expression in tumors has been proven, while eliminating the problem of variability in pathologists' assessments. Artificial intelligence can compensate for staff shortages, reduce physician workload, prevent burnout. The question remains how artificial intelligence will help with complex histomorphological differential diagnosis in the future and how to solve the impact of unrepresentative training data, data confidentiality issues, algorithm fragility, and reduce error rates. Researchers emphasize the indispensable role of human experts in the synthesis of images, clinical data, and experience for complex histopathological diagnostic tasks.