Researchers at King’s College London say they have mapped how the immune response in sepsis changes over time, using machine learning and repeated blood sampling from critically ill patients at Guy’s and St Thomas’ NHS Foundation Trust. The findings, published in Immunity on 28 August 2026, suggest that the immune state of a patient with sepsis may not always match the clinical stage doctors assign at the bedside.
What the study found
The team analyzed blood samples taken at four timepoints between admission to and discharge from critical care. They examined multiple layers of immune information, including immune cells, gene expression and protein expression, to build what they described as an immune profile of sepsis.
According to the researchers, the results showed that different sepsis immune states did not line up neatly with the usual clinical timeline. In one example, the early immune state was not the same as the early clinical stage of sepsis, which is often defined by the day a clinician diagnoses the condition.
The study was led by Dr Matthew Fish during his PhD at King’s College London, and involved Professor Manu Shankar-Hari, Professor Mervyn Singer and other researchers. The authors said the work could help identify which treatments are most likely to work, and when they should be given.
Why timing matters in sepsis care
Sepsis remains a major clinical challenge because it is driven by a misfiring immune response to infection. The research team argued that looking only at a single snapshot may miss important changes that happen as the illness evolves. They said understanding those shifts could help clinicians move toward more precise treatment decisions, rather than relying on broad assumptions about where a patient is in the course of the disease.
Professor Manu Shankar-Hari said the research shows the importance of looking at the changing architecture of the immune system in sepsis over time, rather than taking a snapshot view. The next step, the authors said, is to understand what causes immune changes between infection onset and the development of sepsis, with the aim of identifying new treatment targets and approaches.
What comes next
The study adds to growing interest in using data-driven tools to understand complex critical illness. By combining machine learning with detailed biological sampling, the researchers hope to refine how sepsis is classified and treated, and eventually improve outcomes for patients with serious infections.
The findings may also matter beyond the UK, since sepsis affects millions of people worldwide and current treatment strategies still rely heavily on timely recognition and supportive care. If future studies confirm the pattern seen here, clinicians could one day use immune profiling to guide more personalized interventions in critical care.
Source: News-Medical report and Immunity study.
Photo source: AI-generated or edited image
