FebriDx host response point-of-care testing improves patient triage for coronavirus disease 2019 (COVID-19) in the emergency department.

OBJECTIVES: Patients presenting to hospital with suspected coronavirus disease 2019 (COVID-19), based on clinical symptoms, are routinely placed in a cohort together until polymerase chain reaction (PCR) test results are available. This procedure leads to delays in transfers to definitive areas and high nosocomial transmission rates. FebriDx is a finger-prick point-of-care test (PoCT) that detects an antiviral host response and has a high negative predictive value for COVID-19. We sought to determine the clinical impact of using FebriDx for COVID-19 triage in the emergency department (ED). DESIGN: We undertook a retrospective observational study evaluating the real-world clinical impact of FebriDx as part of an ED COVID-19 triage algorithm. SETTING: Emergency department of a university teaching hospital. PATIENTS: Patients presenting with symptoms suggestive of COVID-19, placed in a cohort in a 'high-risk' area, were tested using FebriDx. Patients without a detectable antiviral host response were then moved to a lower-risk area. RESULTS: Between September 22, 2020, and January 7, 2021, 1,321 patients were tested using FebriDx, and 1,104 (84%) did not have a detectable antiviral host response. Among 1,104 patients, 865 (78%) were moved to a lower-risk area within the ED. The median times spent in a high-risk area were 52 minutes (interquartile range [IQR], 34-92) for FebriDx-negative patients and 203 minutes (IQR, 142-255) for FebriDx-positive patients (difference of -134 minutes; 95% CI, -144 to -122; P < .0001). The negative predictive value of FebriDx for the identification of COVID-19 was 96% (661 of 690; 95% CI, 94%-97%). CONCLUSIONS: FebriDx improved the triage of patients with suspected COVID-19 and reduced the time that severe acute respiratory coronavirus virus 2 (SARS-CoV-2) PCR-negative patients spent in a high-risk area alongside SARS-CoV-2-positive patients.

Using explainable machine learning to characterise data drift and detect emergent health risks for emergency department admissions during COVID-19.

A key task of emergency departments is to promptly identify patients who require hospital admission. Early identification ensures patient safety and aids organisational planning. Supervised machine learning algorithms can use data describing historical episodes to make ahead-of-time predictions of clinical outcomes. Despite this, clinical settings are dynamic environments and the underlying data distributions characterising episodes can change with time (data drift), and so can the relationship between episode characteristics and associated clinical outcomes (concept drift). Practically this means deployed algorithms must be monitored to ensure their safety. We demonstrate how explainable machine learning can be used to monitor data drift, using the COVID-19 pandemic as a severe example. We present a machine learning classifier trained using (pre-COVID-19) data, to identify patients at high risk of admission during an emergency department attendance. We then evaluate our model's performance on attendances occurring pre-pandemic (AUROC of 0.856 with 95%CI [0.852, 0.859]) and during the COVID-19 pandemic (AUROC of 0.826 with 95%CI [0.814, 0.837]). We demonstrate two benefits of explainable machine learning (SHAP) for models deployed in healthcare settings: (1) By tracking the variation in a feature's SHAP value relative to its global importance, a complimentary measure of data drift is found which highlights the need to retrain a predictive model. (2) By observing the relative changes in feature importance emergent health risks can be identified.