A study in Immunity profiled blood at four points from admission to discharge among people in intensive care at Guy's and St Thomas'. Machine learning separated three successive immune states called STImS; the early state does not line up with the early clinical stage. Sepsis is linked worldwide to more than 160 million cases and about 21 million deaths each year. Matthew Fish and Manu Shankar-Hari argue that treatment timing may need to follow which temporal state someone is in.
The Immunity study profiled immune cells, gene expression and protein levels together in people in intensive care at Guy's and St Thomas'. Blood was taken at four points from admission to discharge. News-Medical reports that researchers combined these layers to build a comprehensive immune profile in people with sepsis for the first time; each state involves different immune-cell activity and response programmes.[1], [2]
Machine-learning models split the data into three successive temporal states called STImS. The early immune state does not fall in the same window as the early clinical stage of the illness; clinical and immune stages slip past each other. Matthew Fish and Manu Shankar-Hari argue that whether a drug works may depend on which immune state someone is in rather than on which day of illness. Fish carried out the work during his PhD at King's College London and now continues sepsis research at Massachusetts General Hospital.[1], [2]
Sepsis is linked worldwide to more than 160 million cases and about 21 million deaths each year; current treatments target infection and organ failure, and efforts to correct the misfiring immune response have not yet improved outcomes. The work rests on one hospital cohort, and no trial has yet tested whether choosing treatment by state changes outcomes. Fish and Shankar-Hari call for watching the changing architecture of the immune system over time instead of taking a single snapshot.[1], [2]
