Development, validation and application of population-based pulmonary disease models using robustness analysis and ensemble forecasting

Lung diseases affecting the critically ill (acute respiratory distress syndrome (ARDS), pneumonia and ventilator-associated lung injury) are acute, severe injuries affecting most or all of both lungs. Patients with these conditions experience defects in gas-exchange and tissue oxygenation and often require mechanical ventilation (life support) because of respiratory failure. While medical researchers and physiologists have been studying these disease for many years, very limited progress has been made in understanding the underlying causes and mechanisms (especially of ARDS and VALI). As a result of the explosive developments and demonstrated successes of molecular systems biology in the last decade, many researchers are now attempting to apply similar systems-based computational approaches to develop and analyse physiological simulation models at the organ level. However, the uncertainty that arises due to patient and disease heterogeneity, and the difficulty in rigorously validating simulation models which take this uncertainty into account, represent serious roadblocks to progress in applying systems approaches in a clinical setting. We propose to tackle both of these problems by adopting a highly novel and interdisciplinary approach. Using approaches from the field of Control Engineering, we will develop population-based disease models which explicitly take account of uncertainty and variability both within a single patient and between diverse members of a patient population, and rigorously validate these models using advanced robustness analysis techniques. By adapting ensemble modelling techniques from the field of climate forecasting, we will develop reliable descriptors of disease severity and novel disease-specific therapeutic strategies that are applicable to populations of individual patients, rather than a single "typical" patient. By transferring state-of-the-art technologies from control engineering and climate science into critical-care medicine, this project will deliver a significant improvement in the understanding, diagnosis and treatment of ARDS, pneumonia and VALI, and a corresponding reduction in both their impact on patients and the associated cost to the NHS.

The huge potential for the application of systems approaches and computational modelling in healthcare has so far been limited by the lack of rigorous model validation procedures, and by the difficulty in applying physiological models to heterogeneous patient populations. Our proposal has the potential to directly address both these issues and expand the applicability of systems biology approaches to real clinical situations. The substantial impact offered by the proposal are in main two areas: (i) the advancement of techniques for the development, validation and analysis of computational models in healthcare research, and (ii) advances in the systems level understanding and clinical management of pulmonary disease states:
- Advances in the exploitation of systems engineering / systems biology approaches in healthcare: Validated, high-fidelity, organ-level models are under-used in medical research; they offer many potential insights into real-world clinical scenarios and provide a powerful, credible and cost-effective methodology to investigate different therapeutic strategies in a cost-effective manner. In the proposed project we expect to make significant advances in this area, and to demonstrate the impact of (i) modelling of uncertain parameters representing variations across patient populations (to provide credible outcomes relating to many individual members of the patient population, rather than the traditional and overly-simplistic "average" patient), (ii) development of novel, rigorous and scalable model validation techniques. In particular, the development and application of novel methods for validating pathophysiological simulation models could enable significant breakthroughs in this field. Compared to the state-of-the-art in systems engineering, current approaches to the validation of physiological models are weak and not fit for purpose; we believe that through "smart validation" we can radically reduce the human validation studies required to make the results of computational studies applicable to clinical practice.
- Advances in the delivery of clinical care: An increased understanding of the potential targets for management of pulmonary disease states has the potential to radically improve critical care and acute/emergency medicine. Currently, there is significant clinical controversy about optimal management of patients with pulmonary disease, particularly in the surgical, emergency and critical care medicine setting. The proposed research project will have a direct impact on the delivery of clinical care, since the investigator at Nottingham is an "end-user" of our proposed modelling framework with the ability to test, evaluate and implement new therapeutic strategies arising from our research.