Roimi M, et al. - Utilizing electronic health records of 2,351 patients from Beth Israel Deaconess Medical Center (BIDMC) in Boston, Massachusetts, USA, and 1,021 from Rambam Health Care Campus (RHCC), Haifa, Israel, researchers tried to construct a machine-learning (ML) algorithm that is predictive of intensive care unit (ICU)-acquired bloodstream infections (BSI) in a patient population suspected of infection in the ICU. Adults from whom blood cultures were obtained for suspected BSI at least 48 h following admission were the participants. Using an ML algorithm, they assessed clinical data, including time-series variables and their interactions, at each site. For BIDMC and for RHCC, the means of the cross-validation AUROCs were estimated to be 0.87 ± 0.02 and 0.93 ± 0.03, respectively. High performance in identifying BC samples with a high likelihood of ICU-acquired BSI was achieved with an ML approach that utilizes temporal and site-specific data.