Developing and validating a machine learning prognostic model for alerting to imminent deterioration of hospitalized patients with COVID-19.

Our study was aimed at developing and validating a new approach, embodied in a machine learning-based model, for sequentially monitoring hospitalized COVID-19 patients and directing professional attention to patients whose deterioration is imminent. Model development employed real-world patient data (598 prediction events for 210 patients), internal validation (315 prediction events for 97 patients), and external validation (1373 prediction events for 307 patients). Results show significant divergence in longitudinal values of eight routinely collected blood parameters appearing several days before deterioration. Our model uses these signals to predict the personal likelihood of transition from non-severe to severe status within well-specified short time windows. Internal validation of the model's prediction accuracy showed ROC AUC of 0.8 and 0.79 for prediction scopes of 48 or 96 h, respectively; external validation showed ROC AUC of 0.7 and 0.73 for the same prediction scopes. Results indicate the feasibility of predicting the forthcoming deterioration of non-severe COVID-19 patients by eight routinely collected blood parameters, including neutrophil, lymphocyte, monocyte, and platelets counts, neutrophil-to-lymphocyte ratio, CRP, LDH, and D-dimer. A prospective clinical study and an impact assessment will allow implementation of this model in the clinic to improve care, streamline resources and ease hospital burden by timely focusing the medical attention on potentially deteriorating patients.

Red blood cell transfusions--are we narrowing the evidence-practice gap? An observational study in 5 Israeli intensive care units.

PURPOSE: The aim of the study was to document transfusion practices in a cross section of general intensive care units (ICUs) in Israel and to determine whether current guidelines are being applied. MATERIALS AND METHODS: This prospective study was performed in 5 general ICUs in Israel over a 3-month period. Red cell transfusion data collected on consecutive patients included the trigger, units transfused per transfusion event, and indications, categorized either to treat a specified condition for which transfusions may be beneficial (acute hemorrhage, acute myocardial ischemia, or severe sepsis) or to treat a low hemoglobin concentration. RESULTS: Of the 238 patients studied, 50% received at least one red blood cell transfusion. The main indication for transfusion (43.7%, or 162/368 U transfused) was to treat a low hemoglobin concentration, in the absence of one of the specified conditions. Total red cell use was 3.0 ± 2.9 U per admission, and patients received a mean of 1.2 ± 0.4 U per transfusion event. The transfusion trigger for the whole group was 7.9 ± 1.1 g/dL. This did not differ significantly between the indications apart from a significantly higher trigger for patients with acute myocardial ischemia (8.8 ± 0.9 g/dL). In addition, patients with a history of heart disease had a higher trigger irrespective of the primary indication for transfusion and received significantly more units per transfusion event. Patients receiving a transfusion had significantly longer ICU stay and hospital mortality. CONCLUSIONS: Our study showed that evidence-practice gaps continue to exist, and it appears that physician behavior is mainly driven by the absolute level of hemoglobin. Educational interventions focused on these factors are required to limit the widespread and often unnecessary use of this scarce and potentially harmful resource.