Title Overview of clinical validation processes for artificial intelligence applications in pathology /
Translation of Title Overview of Clinical Validation Processes for Artificial Intelligence Applications in Pathology.
Authors Lyzogub, Margaryta
Full Text Download
Pages 28
Keywords [eng] Artificial intelligence, machine learning, deep learning, supervised learning, computational pathology, validation, Ki-67 enumeration algorithm
Abstract [eng] The paper aims to provide a general understanding of the impact of artificial intelligence on the current pathology and explore the aspects that a medical student or a clinician without a background in the computational pathology need to be aware of to evaluate the performance of the algorithm. It starts with the overview of the history of computational pathology development emphasizing the speed and the effort needed for the novel trends’ introduction into research and clinical practice. In then explores the most common types of algorithms deployed in the industry and cases of their application. An idealistic model of fully symbiotic machine-pathologist workload is depicted. Next, information about the process of artificial intelligence models’ development is introduced, focusing on the steps of analytical and clinical validation. The approaches to the data distribution for training, tuning, and testing are discussed and the most commonly used statistical measures are explained. The work concludes with the ideas on the steps needed to take today in terms of medical education, to provide the healthcare specialists of tomorrow with the relevant knowledge to face the era or artificial intelligence prepared. In conclusion, it takes a combination of pathology, data science, medical statistics, medical law and many more areas of knowledge to accurately assess the novel model, thus providing the healthcare professionals with basic concepts in diverse expertise domains during medical education using stratified approach based on their future role in AI development is crucial.
Dissertation Institution Vilniaus universitetas.
Type Master thesis
Language English
Publication date 2024