Large randomized clinical trials of immunomodulatory interventions for acute inflammatory diseases such as sepsis have had a dismal track record. The biological complexity of the host-pathogen interaction and the potential large impact of a successful treatment on the health care system and society position diseases such as sepsis as ideal test beds for model-based therapeutic approaches, as proposed in the FDA critical path document and the NIH roadmap initiative. Yet, there is a paucity of organism-level computational models of inflammation. More fundamental however, is the lack of human data sets where such models could be validated. Such a data set would be extraordinarily expensive to assemble and is highly unlikely to be acquired merely for testing model-based interventions in the absence of models with demonstrated validity. The NIH-funded Protocolized Care for Early Septic Shock (ProCESS) study is currently examining the impact of early resuscitation in victims of severe sepsis in a 1350 patient prospective randomized trial and will produce a data set with a granularity that will not only help to understand the processes involved in sepsis, but also the biological consequences of a physiologic goal-directed treatment protocol. The overarching goal of the program outlined in this proposal is to validate computational models of human sepsis using data from the ProCESS study through advanced mathematical and computational methods. We have assembled a transdisciplinary group of modelers and clinicians with an eloquent track record of successful collaboration on developing, calibrating and testing in silico models of acute inflammation, and of sepsis in particular, of different levels of granularity. We believe that validation of in silico models in a large clinically relevant cohort is absolutely crucial to the legitimization of computational modeling as a technology that will prove pivotal to the design of smarter randomized interventional trials in general, and of personalized therapies in particular. Leveraging data and preliminary analyses from the ProCESS trial on the one hand and an extensive existing transdisciplinary effort at expanding existing computational models of the acute inflammatory response on the other will also provide an unprecedented opportunity to gain mechanistic understanding of the processes leading to organ failure and death, systemic recovery and unexpected failure.
Computational modeling has been an integral part of knowledge discovery in many fields, but not clinical medicine, owing mostly to the complexity of the human as a system and the absence of adequate data sets where models could be validated. Yet, computational models will prove central to the design of smart and individualized treatment strategies of complex diseases. We are proposing to develop such practical models for the specific case of sepsis, a condition which has proven particularly difficult to treat and where clinical trials of new drugs have failed repeatedly. Combining the expertise of a transdisciplinary group of modelers pioneering infection research and the vast data set generated from a prospective, randomized trial controlled trial of how to best support patients in the first few hours of sepsis, we will construct and validate predictive computational models of sepsis which could form the basis for smarter clinical trials and personalized therapies.
|Zhang, Li Ang; Parker, Robert S; Swigon, David et al. (2016) A One-Nearest-Neighbor Approach to Identify the Original Time of Infection Using Censored Baboon Sepsis Data. Crit Care Med 44:e432-42|
|Malkin, Alexander D; Sheehan, Robert P; Mathew, Shibin et al. (2015) A Neutrophil Phenotype Model for Extracorporeal Treatment of Sepsis. PLoS Comput Biol 11:e1004314|
|Price, Ian; Mochan-Keef, Ericka D; Swigon, David et al. (2015) The inflammatory response to influenza A virus (H1N1): An experimental and mathematical study. J Theor Biol 374:83-93|
|Mochan-Keef, Ericka; Swigon, David; Ermentrout, G Bard et al. (2015) A Three-Tiered Study of Differences in Murine Intrahost Immune Response to Multiple Pneumococcal Strains. PLoS One 10:e0134012|