In silico database of ~500 000 virtual patients with diverse cardiovascular disease

The presented dataset contains a large virtual cohort of more than 50'000 heart failure patients characterized by realistic traces of volumes, pressures, flows, and regional mechanics. We used the well-established CircAdapt model (https://www.circadapt.org/ ,  http://framework.circadapt.org/) of the human heart and circulation to simulate a large cohort of virtual patients, covering a wide range of HF-related disease heterogeneity and severity. In the construction of virtual patients, generating a variety of parameter sets for the initial population is essential and can be accomplished using various sampling techniques. In our investigation, we utilized the Sobol-low discrepancy sequence to ensure uniformity across the high-dimensional parameter space. For a more detailed explanation please refer to the document "MARCIUS_deliverable_D1.5_VirtualDatabase.pdf"

Acknowledgments: This work was supported by the European Union's Horizon 2020 Research and Innovation program under the Marie Skłodowska-Curie grant agreement No. 86074, "MARCIUS - MARie Curie Intelligent UltraSound"(https://www.marcius-project.com/). MARCIUS rationale was to develop a comprehensive in silico simulation platform comprising both the generation of virtual patients (presented dataset) and their associated realistic image data. Such approach would make it possible to learn the most relevant patterns within a wide representative set of patients to lead the training of ML-based image processing algorithms in order to analyze real-world clinical data.