Smart Early Recognition of Sepsis
University Medical Center Groningen and the private sector partner Inflammatix, Inc. join forces to improve diagnosis and clinical decision support in patients with early sepsis.
Before the pandemic, 50 million patients were annually affected by sepsis worldwide. The syndrome is caused by the body’s own dysregulated response to a bacterial, viral, or fungal infection. Because of the maladaptive process, various organs of the patient can be damaged, causing 11 million sepsis-related deaths worldwide. The pandemic with severe cases of COVID19 meeting the definition of viral sepsis has led to a steep increase in these numbers. The superior goal in healthcare is to recognise infection as early as possible, distinguishing viral from bacterial infections to inform therapy. The latter requires urgent application of antibiotics, with each hour of delayed treatment introducing additional risk for the patient. But those treatments come with a flipside: rampant overuse of antibiotics drives the development of antimicrobial-resistant bacteria, rendering the currently available antibiotics useless. The newest estimates indicate nearly 5 million deaths associated with antimicrobial resistance. Novel precision diagnostic approaches can help balance between overtreating the uninfected and missing the infected patients, thereby improving individual patient outcomes and reducing the socioeconomic burden of sepsis and AMR on society.
The project will approach the diagnostic dilemma by integrating various information sources – patient demography and history, vital signs (e.g., heart rate) as well as laboratory values (e.g., number of immune cells in blood), into the development of a single decision support tool for diagnosing, prognostication as well as the identification of treatment response. The project will leverage the existing Acutelines framework within the emergency department of the UMC Groningen, which ensures availability and standardised processing of data and biomaterial such as blood. Overall, high-dimensional data from 1,600 patients will drive the research project. Molecular signatures from the blood – derived from both the activity (expression) of genes in immune cells and the proteins available in the liquid part of the blood – will constitute the core layer of information. Therefore, large parts of the project will aim to identify and validate novel and existing molecular signatures and bring them into context with other data. This approach requires the use of state-of-the-art machine learning algorithms, a core expertise Inflammatix brings into the collaboration, besides the technical proficiency of providing rapid point-of-care-based gene expression diagnostics.
The project aims to deliver better clinical decision support for patients with suspected sepsis – ranging from molecular signature discovery, validation, data integration to reaching a health economic and outcome evaluation and prospective performance validation of the final machine-learning enabled model supported by rapid point-of-care gene expression analysis.
