Machine learning supported multi-model simulator for infection research - MuMoSim

Abstract

The Machine Learning Supported Multi-Model Simulator for Infection Research (MuMoSim) project aims to advance modeling in infection research by developing a next-generation quantitative and predictive simulation tool. While recent efforts in biomathematical modeling have primarily focused on implementing spatio-temporal multiscale models, the MuMoSim framework introduces a new dimension in biomedical system simulation, enhanced by modern machine learning methods. It adopts a multi-model approach to consistently integrate various modeling strategies, each reflecting different levels of complexity within the biomedical system, into a unified framework. The multi-model simulator supports bottom-up modeling – beginning with approaches that quantitatively describe existing experimental data and progressing to higher-level models that generate testable predictions for subsequent experimental validation. It also enables the stepwise integration of diverse experimental data types, including time-resolved flow cytometry, survival assays, and microscopic imaging. The project's primary focus is the quantitative and predictive modeling of bloodstream infections caused by microbial pathogens, using experimental data from human whole-blood infection assays (WBIA), with general applicability of the simulator to interacting cells and molecular systems in biomedicine.