Wright SW, Kaewarpai T, Lovelace-Macon L, et al. - Given the value of identification of melioidosis patients at high risk of clinical deterioration for guiding decisions about resource allocation and management, researchers attempted to develop a biomarker-based model for 28-day mortality prediction in melioidosis. They prospectively assessed 113 hospitalized Thai patients with melioidosis (a derivation set) for the concentrations of interferon-γ, interleukin-1β, interleukin-6, interleukin-8, interleukin-10, tumor necrosis factor-α, granulocyte-colony stimulating factor, and interleukin-17A. A subset of predictive biomarkers was identified using least absolute shrinkage and selection operator (LASSO) regression and biomarker-based prediction of 28-day mortality was evaluated compared with clinical variables via performing logistic regression and receiver operating characteristic curve analysis. LASSO regression led to the selection of interleukin-6 and interleukin-8 among the eight cytokines that were positively associated with 28-day mortality. Significant improvement in 28-day mortality prediction was observed with a model consisting of interleukin-6, interleukin-8 and clinical variables vs a model of only clinical variables. In both internal validation set (N = 78) and in a prospectively enrolled external validation set (N = 161) of hospitalized adults with melioidosis in Thailand, the combined model including biomarkers led to significant improvement in 28-day mortality prediction over a model limited to clinical variables.