Precision medicine in sepsis: from biological classification to treatment decisions

Precision medicine in sepsis should be organised around revisable treatment decisions. Molecular endotypes and immune biomarkers have advanced our understanding of heterogeneity, but a classification associated with mortality does not necessarily identify who benefits from a particular therapy. We argue that clinical translation requires joint interpretation of infection attribution, immune competence, tissue dysfunction and their evolution over time, informed by clinical judgement. Recent advances in immune profiling, biomarker-selected immunotherapy and personalised resuscitation make this approach increasingly testable. Their relevance lies in the decisions they enable, while their limitations show why biological plausibility, target engagement and patient benefit must remain distinct. We propose that future studies evaluate complete strategies linking measurement to interpretation, treatment and reassessment, including explicit conditions for changing or stopping therapy. Multi-omics should help identify mechanisms and clinically usable measurements; artificial intelligence should integrate longitudinal observations and make disagreement between clinical assessment and measured physiology visible. Neither should displace timely antimicrobial treatment, source control or bedside responsibility. This Perspective sets out a research agenda in which precision is demonstrated by better treatment choices and patient-centred outcomes at defined stages of illness.