Pre-hospital prediction of adverse outcomes in patients with suspected COVID-19: Development, application and comparison of machine learning and deep learning methods.

BACKGROUND: COVID-19 infected millions of people and increased mortality worldwide. Patients with suspected COVID-19 utilised emergency medical services (EMS) and attended emergency departments, resulting in increased pressures and waiting times. Rapid and accurate decision-making is required to identify patients at high-risk of clinical deterioration following COVID-19 infection, whilst also avoiding unnecessary hospital admissions. Our study aimed to develop artificial intelligence models to predict adverse outcomes in suspected COVID-19 patients attended by EMS clinicians. METHOD: Linked ambulance service data were obtained for 7,549 adult patients with suspected COVID-19 infection attended by EMS clinicians in the Yorkshire and Humber region (England) from 18-03-2020 to 29-06-2020. We used support vector machines (SVM), extreme gradient boosting, artificial neural network (ANN) models, ensemble learning methods and logistic regression to predict the primary outcome (death or need for organ support within 30 days). Models were compared with two baselines: the decision made by EMS clinicians to convey patients to hospital, and the PRIEST clinical severity score. RESULTS: Of the 7,549 patients attended by EMS clinicians, 1,330 (17.6%) experienced the primary outcome. Machine Learning methods showed slight improvements in sensitivity over baseline results. Further improvements were obtained using stacking ensemble methods, the best geometric mean (GM) results were obtained using SVM and ANN as base learners when maximising sensitivity and specificity. CONCLUSIONS: These methods could potentially reduce the numbers of patients conveyed to hospital without a concomitant increase in adverse outcomes. Further work is required to test the models externally and develop an automated system for use in clinical settings.

Prognostic accuracy of triage tools for adults with suspected COVID-19 in a prehospital setting: An observational cohort study

Aims/Objectives/Background Emergency Medical Service (EMS) and other practitioners assessing patients with suspected COVID-19 in the community must rapidly determine whether patients need treatment in hospital or can self-care. Tools to triage patient acuity have only been validated in hospital populations. We aimed to estimate the accuracy of five risk-stratification tools recommended to predict severe illness and compare accuracy to existing clinical decision-making in a pre-hospital setting. Methods/Design An observational cohort study using linked ambulance service data for patients assessed by EMS crews in the Yorkshire and Humber region of England between 18th March 2020 and 29th June 2020 was conducted to assess performance of the PRIEST tool, NEWS2, the WHO algorithm, CRB-65 and PMEWS in patients with suspected COVID-19 infection. The primary outcome was death or need for organ support. Results/Conclusions Of 7550 patients in our cohort, 17.6% (95% CI:16.8% to 18.5%) experienced the primary outcome. The NEWS2, PMEWS, PRIEST tool and WHO algorithm identified patients at risk of adverse outcomes with a high sensitivity (>0.95) and specificity ranging between 0.3 (NEWS2) and 0.41 (PRIEST tool). The high sensitivity of NEWS2 and PMEWS was achieved by using lower thresholds than previously recommended (NEWS2;0-1 vs 2+ and PMEWS;0-2 vs 3+). On index (first) assessment, 65% of patients were transported to hospital and EMS decision to transfer patients achieved a sensitivity of 0.84 (95% CI 0.83 to 0.85) and specificity of 0.39 (95% CI 0.39 to 0.40) to the primary outcome. This does not account for clinical reasons not to convey patients to hospital who subsequently deteriorated. Use of NEWS2, PMEWS, PRIEST tool and WHO algorithm could therefore potentially improve EMS triage of patients with suspected COVID-19 infection. Use of the PRIEST tool could significantly increase the sensitivity of triage without increasing the number of patients conveyed to hospital. (Table Presented).