In the last 10 years there has been great progress in advancing multiscale numerical methods for modeling biological systems in diverse projects sponsored by MSM-IMAG, e.g. for cancer, hematological disorders, cardiovascular diseases, etc. The main theme has been the tight coupling between heterogeneous methods across two different scale regimes, e.g. continuum (macro) and atomistic (micro). More recently, uncertainty quantification (UQ), global sensitivity, and estimation of parameters under uncertainty have emerged as enabling methods for the quantitative probing of multiscale biological systems and for the construction of predictive models. In this presentation, we will review some of these methodologies and corresponding software, and will introduce new concepts emphasizing Bayesian modeling, multi-fidelity approaches and propagation of uncertainty. Multi-fidelity modeling allows the synergistic use of diverse models across scales, relying on a few high-fidelity data or simulations and many more inexpensive but less accurate measurements or simulations. This efficient framework naturally integrates simulations and data while at the same time predicts the total uncertainty of the simulated system, i.e. model and parametric uncertainty or even numerical inaccuracies for multi-resolution runs. It also resembles active learning and it naturally leads to the selection of the next best experiment or simulation in order to reduce the uncertainty in the predictions. These concepts will further be developed by the invited speakers in this session who will elaborate more on e.g. Bayesian modeling, deep learning, data inference, etc.