Prof. Milos Kojic

Abstract

We are witnessing that the era of Artificial Intelligence (AI) has arrived. AI is used in
everyday life in all conversations, starting from science to decision-making, to social life. It is of particular interest in medicine, where the hope is that AI as a tool will help in the understanding of diseases as well as in the improvement of medical treatment outcomes. AI in medical research and practice may rely on various types of models, such as statistical, empirical, physics-based, or others. In this presentation, we consider physics-based computational models in biomedical engineering and emphasize the main features of these models to be used as suitable tools within AI. Generally, these models should be robust, accurate, and efficient. Our group in Kragujevac, Serbia, has experience of more than five decades of development of our finite element program PAK with applications in engineering and biomedical engineering. We here outline our methodology in biomedical engineering developed to achieve the mentioned model characteristics. The selected examples include the modeling of motion of deformable bodies (as cells) within a fluid which is based on the strong coupling and remeshing procedure; mass transport and electrophysiology within composite media as tissue, according to our general smeared multiscale-multiphysics concept (Kojic Transport Model) with coupling different physical fields; and solid-fluid interaction. Applications of these models are illustrated in drug delivery with tumor growth; the role of platelets in metastasis; heart mechanics and electrophysiology; lung mechanics, airflow, blood flow, and gas exchange. In conclusion regarding the presented methodologies and applications within AI, we emphasize that model efficiency can be considered of secondary importance since today and future technologies provide larger computational speed, parallel computing, generation of huge number and surrogate model solutions, machine learning, and advanced graphical support.