Project Details
Projekt
P5: Statistical methods for data integration into multi-scale models and uncertainty tracking
Applicant
Professorin Dr. Nicole Radde
Subject Area
General and Visceral Surgery
Term
since 2020
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 436883643
The goal of QuaLiPerF is an advanced understanding of the interplay between surgery-induced changes in perfusion and metabolic function for improving predictions of functions and risk stratification for patients subjected to (extended) liver resection (ePHx). This will be achieved by a systems medicine approach. Based on our previous work, we will build spatially and temporally resolved multi-scale models of hepatic function, perfusion and liver regeneration. Therefore, a variety of data is generated during the course of the project or is available from other studies. The integration of experimental data into these models is a challenging task, which requires multiple decisions during the modeling and calibration procedure. It provides methodology for smart data integration and uncertainty quantification across the modeling project. Furthermore, it develops methods to integrate patient-specific factors in order to enable individualized risk assessment in the context of liver surgery and liver regeneration after surgery. For this purpose, we will use statistical Bayesian and likelihood-based approaches for parameter estimation, uncertainty quantification and model validation. A challenge thereby will be computational costs especially for multi-scale models, since these methods often require many forward simulations. Here, we will develop methodology for up-scaling by exploiting efficient sampling schemes and model reduction techniques. Patient-specific clinical data will be provided by P9, from which features will be selected via correlations with regenerative capacity and post-operative risk assessment. Subsequently, we will develop a scheme to integrate these features into patient-specific models.Methodology will equip model predictions with confidence bounds, thereby judging the reliability and specificity of model predictions. This will not only allow analyzing model plausibility and model validation, but will also be useful in a second funding phase to foster an even tighter experiment-modeling cycle. Ultimately, QuaLiPerF will allow an improved model-based prediction of the recovery of the liver function, thus helping to improve planning of individual surgeries in the future.
DFG Programme
Research Units
