Background: Early risk stratification tools for COVID-19 patients were indicated but little is known about their ability to effectively detect clinical deterioration among COVID-19 patients. Objectives: To assess the prognostic accuracy of NEWS2 in predicting severe morbidity and mortality in hospitalization for COVID-19 patients. Methods: A prospective cohort study was performed among adult patients with a confirmed diagnosis of COVID-19 who were admitted to the inpatient units at COVID-19 Martyrs Medical Military Complex and Palestinian Red Crescent Society Hospital, located at Nablus, Palestinian Authority. NEWS2 scores were measured at admission, on 6, 12, 24, and 48 hours after their admission to the hospital using Receiver Operating Characteristic Curves (ROC). Results: Overall, 192 adult COVID-19 patients aged 25-94 years (mean=62.1, SD=13.9) were enrolled. Of those, 49.0% were males, 47.4% were vaccinated, and 53.6% were with diabetes. The mean NEWS scores during the patient’s hospital stay were excellent predictors of mechanical ventilation, admission to the ICU, and mortality, as indicated by an AUROC of 0.94 (95% CI: 0.88-1.00, P<0.001), 0.91 (95% CI: 0.87-0.96, P<0.001) and 0.96 (95% CI: 0.92-1.00, P<0.001), respectively. The classification of 192 patients into NEWS score categories indicated that 12.5% were classified as high-risk at admission. Conclusion: The NEWS2 had moderate sensitivity and specificity in predicting the deterioration of patients with COVID-19. Our findings support the utility of NEWS2 monitoring as a sensitive approach for initially assessing COVID-19 patients. It could be helpful to improve predictive performance by supplementing the score parameters with additional measures besides the clinical judgment.

This paper presents the development of machine learning (ML) models to predict hypoxemia severity during emergency triage, especially in Chemical, Biological, Radiological, Nuclear, and Explosive (CBRNE) events, using physiological data from medical-grade sensors. Gradient Boosting Models (XGBoost, LightGBM, CatBoost) and sequential models (LSTM, GRU) were trained on physiological and demographic data from the MIMIC-III and IV datasets. A robust preprocessing pipeline addressed missing data, class imbalances, and incorporated synthetic data flagged with masks. Gradient Boosting Models (GBMs) outperformed sequential models in terms of training speed, interpretability, and reliability, making them well-suited for real-time decision-making. While their performance was comparable to that of sequential models, the GBMs used score features from six physiological variables derived from the enhanced National Early Warning Score (NEWS) 2, which we termed NEWS2+. This approach significantly improved prediction accuracy. While sequential models handled temporal data well, their performance gains didn’t justify the higher computational cost. A 5-minute prediction window was chosen for timely intervention, with minute-level interpolations standardizing the data. Feature importance analysis highlighted the significant role of mask and score features in enhancing both transparency and performance. Temporal dependencies proved to be less critical, as Gradient Boosting Models were able to capture key patterns effectively without relying on them. This study highlights ML's potential to improve triage and reduce alarm fatigue. Future work will integrate data from multiple hospitals to enhance model generalizability across clinical settings.