Fig. 5: Ideal threshold for immune subgrouping identification depends on clinical cohort and question.

a, Mean and s.d. of the myeloid (x axis) and lymphoid (y axis) dysregulation scores represented graphically, stratified by severity (defined by critical versus noncritical illness). The individual dots represent patient samples colored by severity and circles the mean and s.d. of lymphoid dysregulation scores. Together this shows the differences that patient cohort composition may have on endotype-defining gene expression signatures. b, Evaluation of the proportion of severity subgroups within the quintiles of myeloid dysregulation scores, indicating how different thresholds would affect the patient population above each threshold. c, Evaluation of the proportion of severity subgroups within the quintiles of lymphoid dysregulation scores, indicating how different thresholds would affect the patient population above each threshold. d, Evaluation of the sensitivity (solid line) and specificity (dashed line) of different myeloid dysregulation score thresholds when identifying mild or moderate infections from healthy participants (blue), severe from mild or moderate infections (yellow) and fatal from nonfatal (red) cases. These results show that the ideal dysregulation threshold will depend on the clinical question. e, Evaluation of the sensitivity (solid line) and specificity (dashed line) of different lymphoid dysregulation score thresholds when identifying mild or moderate infections from healthy participants (blue), severe from mild or moderate infections (yellow) and fatal from nonfatal (red) cases. These results show that the ideal dysregulation threshold will depend on the clinical question.