Scalable federated learning for emergency care using low cost microcomputing: Real-world, privacy preserving development and evaluation of a COVID-19 screening test in UK hospitals (preprint)

Background Tackling biases in medical artificial intelligence requires multi-centre collaboration, however, ethical, legal and entrustment considerations may restrict providers' ability to participate. Federated learning (FL) may eliminate the need for data sharing by allowing algorithm development across multiple hospitals without data transfer. Previously, we have shown an AI-driven screening solution for COVID-19 in emergency departments using clinical data routinely available within 1h of arrival to hospital (vital signs & blood tests; CURIAL-Lab). Here, we aimed to extend and federate our COVID-19 screening test, demonstrating development and evaluation of a rapidly scalable and user-friendly FL solution across 4 UK hospital groups. Methods We supplied a Raspberry Pi 4 Model B device, preloaded with our end-to-end FL pipeline, to 4 NHS hospital groups or their locally-linked research university (Oxford University Hospitals/University of Oxford (OUH), University Hospitals Birmingham/University of Birmingham (UHB), Bedfordshire Hospitals (BH) and Portsmouth Hospitals University (PUH) NHS trusts). OUH, PUH and UHB participated in federated training and calibration, training a deep neural network (DNN) and logistic regressor to predict COVID-19 status using clinical data for pre- pandemic (COVID-19-negative) admissions and COVID-19-positive cases from the first wave. We performed federated prospective evaluation at PUH & OUH, and external evaluation at BH, evaluating the resultant global and site-tuned models for admissions to the respective sites during the second pandemic wave. Removable microSD storage was destroyed on study completion. Findings Routinely collected clinical data from a total 130,941 patients (1,772 COVID-19 positive) across three hospital groups were included in federated training. OUH, PUH and BH participated in prospective federated evaluation, with sets comprising 32,986 patient admissions (3,549 positive) during the second pandemic wave. Federated training improved DNN performance by a mean of 27.6% in terms of AUROC when compared to models trained locally, from AUROC of 0.574 & 0.622 at OUH & PUH to 0.872 & 0.876 for the federated global model. Performance improvement was more modest for a logistic regressor with a mean AUROC increase of 13.9%. During federated external evaluation at BH, the global DNN model achieved an AUROC of 0.917 (0.893-0.942), with 89.7% sensitivity (83.6-93.6) and 76.7% specificity (73.9- 79.1). Site-personalisation of the global model did not give a significant improvement in overall performance (AUROC improvement <0.01), suggesting high generalisability. Interpretations We present a rapidly scalable hardware and software FL solution, developing a COVID-19 screening test across four UK hospital groups using inexpensive micro- computing hardware. Federation improved model performance and generalisability, and shows promise as an enabling technology for deep learning in healthcare. Funding University of Oxford Medical & Life Sciences Translational Fund/Wellcome

Real-world evaluation of AI driven COVID-19 triage for emergency admissions: External validation & operational assessment of lab-free and high-throughput screening solutions (preprint)

BackgroundUncertainty in patients COVID-19 status contributes to treatment delays, nosocomial transmission, and operational pressures in hospitals. However, typical turnaround times for batch-processed laboratory PCR tests remain 12-24h. Although rapid antigen lateral flow testing (LFD) has been widely adopted in UK emergency care settings, sensitivity is limited. We recently demonstrated that AI-driven triage (CURIAL-1.0) allows high-throughput COVID-19 screening using clinical data routinely available within 1h of arrival to hospital. Here we aimed to determine operational and safety improvements over standard-care, performing external/prospective evaluation across four NHS trusts with updated algorithms optimised for generalisability and speed, and deploying a novel lab-free screening pathway in a UK emergency department. MethodsWe rationalised predictors in CURIAL-1.0 to optimise separately for generalisability and speed, developing CURIAL-Lab with vital signs and routine laboratory blood predictors (FBC, U&E, LFT, CRP) and CURIAL-Rapide with vital signs and FBC alone. Models were calibrated during training to 90% sensitivity and validated externally for unscheduled admissions to Portsmouth University Hospitals, University Hospitals Birmingham and Bedfordshire Hospitals NHS trusts, and prospectively during the second-wave of the UK COVID-19 epidemic at Oxford University Hospitals (OUH). Predictions were generated using first-performed blood tests and vital signs and compared against confirmatory viral nucleic acid testing. Next, we retrospectively evaluated a novel clinical pathway triaging patients to COVID-19-suspected clinical areas where either model prediction or LFD results were positive, comparing sensitivity and NPV with LFD results alone. Lastly, we deployed CURIAL-Rapide alongside an approved point-of-care FBC analyser (OLO; SightDiagnostics, Israel) to provide lab-free COVID-19 screening in the John Radcliffe Hospitals Emergency Department (Oxford, UK), as trust-approved service improvement. Our primary improvement outcome was time-to-result availability; secondary outcomes were sensitivity, specificity, PPV, and NPV assessed against a PCR reference standard. We compared CURIAL-Rapides performance with clinician triage and LFD results within standard-care. Results72,223 patients met eligibility criteria across external and prospective validation sites. Model performance was consistent across trusts (CURIAL-Lab: AUROCs range 0.858-0.881; CURIAL-Rapide 0.836-0.854), with highest sensitivity achieved at Portsmouth University Hospitals (CURIAL-Lab:84.1% [95% Wilsons score CIs 82.5-85.7]; CURIAL-Rapide:83.5% [81.8 - 85.1]) at specificities of 71.3% (95% Wilsons score CIs: 70.9 - 71.8) and 63.6% (63.1 - 64.1). For 3,207 patients receiving LFD-triage within routine care for OUH admissions between December 23, 2021 and March 6, 2021, a combined clinical pathway increased sensitivity from 56.9% for LFDs alone (95% CI 51.7-62.0) to 88.2% with CURIAL-Rapide (84.4-91.1; AUROC 0.919) and 85.6% with CURIAL-Lab (81.6-88.9; AUROC 0.925). 520 patients were prospectively enrolled for point-of-care FBC analysis between February 18, 2021 and May 10, 2021, of whom 436 received confirmatory PCR testing within routine care and 10 (2.3%) tested positive. Median time from patient arrival to availability of CURIAL-Rapide result was 45:00 min (32-64), 16 minutes (26.3%) sooner than LFD results (61:00 min, 37-99; log-rank p<0.0001), and 6:52 h (90.2%) sooner than PCR results (7:37 h, 6:05-15:39; p<0.0001). Sensitivity and specificity of CURIAL-Rapide were 87.5% (52.9-97.8) and 85.4% (81.3-88.7), therefore achieving high NPV (99.7%, 98.2-99.9). CURIAL-Rapide correctly excluded COVID-19 for 58.5% of negative patients who were triaged by a clinician to COVID-19-suspected (amber) areas. ImpactCURIAL-Lab & CURIAL-Rapide are generalisable, high-throughput screening tests for COVID-19, rapidly excluding the illness with higher NPV than LFDs. CURIAL-Rapide can be used in combination with near-patient FBC analysis for rapid, lab-free screening, and may reduce the number of COVID-19-negative patients triaged to enhanced precautions ( amber) clinical areas.

Altered neutrophil phenotype and function in non-ICU COVID-19 patients correlated with disease severity (preprint)

Rational: Infection with the SARS-CoV2 virus is associated with elevated neutrophil counts. Evidence of neutrophil dysfunction in COVID-19 is based predominantly on transcriptomics or single functional assays. Cell functions are interwoven pathways, and so understanding the effect of COVID-19 across the spectrum of neutrophil function may identify tractable therapeutic targets. Objectives: Examine neutrophil phenotype and functional capacity in COVID-19 patients versus age-matched controls (AMC) Methods: Isolated neutrophils from 41 hospitalised, non-ICU COVID-19 patients and 23 AMC underwent ex vivo analyses for migration, bacterial phagocytosis, ROS generation, NET formation (NETosis) and cell surface receptor expression. DNAse 1 activity was measured, alongside circulating levels of cfDNA, MPO, VEGF, IL-6 and sTNFRI. All measurements were correlated to clinical outcome. Serial sampling on day 3-5 post hospitalisation were also measured. Results: Compared to AMC, COVID-19 neutrophils demonstrated elevated transmigration (p=0.0397) and NETosis (p=0.0366), but impaired phagocytosis (p=0.0236) associated with impaired ROS generation (p<0.0001). Surface expression of CD54 (p<0.0001) and CD11c (p=0.0008) was significantly increased and CD11b significantly decreased (p=0.0229) on COVID-19 patient neutrophils. COVID-19 patients showed increased systemic markers of NETosis including increased cfDNA (p=0.0153) and impaired DNAse activity (p<0.0.001). MPO (p<0.0001), VEGF (p<0.0001), TNFRI (p<0.0001) and IL-6 (p=0.009) were elevated in COVID-19, which positively correlated with disease severity by 4C score. Conclusion: COVID-19 is associated with neutrophil dysfunction across all main effector functions, with altered phenotype, elevated migration, impaired antimicrobial responses and elevated NETosis. These changes represent a clear mechanism for tissue damage and highlight that targeting neutrophil function may help modulate COVID-19 severity.

CATALYST trial protocol: A multicentre, open-label, phase II, multi-arm trial for an early and accelerated evaluation of the potential treatments for COVID-19 in hospitalised adults (preprint)

IntroductionSevere SARS-CoV-2 infection is associated with a dysregulated immune response. Inflammatory monocytes and macrophages are crucial, promoting injurious, pro-inflammatory sequelae. Immunomodulation is, therefore, an attractive therapeutic strategy and we sought to test licensed and novel candidate drugs. Methods and analysisThe CATALYST trial is a multi-arm, open-label, multi-centre, phase II platform trial designed to identify candidate novel treatments to improve outcomes of patients hospitalised with COVID-19 compared with usual care. Treatments with evidence of biomarker improvements will be put forward for larger-scale testing by current national phase III platform trials. Hospitalised patients >16 years with a clinical picture strongly suggestive of SARS-CoV-2 pneumonia (confirmed by chest X-ray or CT scan, with or without a positive reverse transcription polymerase chain reaction (RT-PCR) assay) and a C-Reactive Protein (CRP) [≥]40 mg/L are eligible. The primary outcome measure is CRP, measured serially from admission to day 14, hospital discharge or death. Secondary outcomes include the WHO Clinical Progression Improvement Scale as a principal efficacy assessment. Ethics and disseminationThe protocol was approved by the East Midlands - Nottingham 2 Research Ethics Committee (20/EM/0115) and given Urgent Public Health status; initial approval was received on 05-May-2020, current protocol version (v6.0) approval on 12-Oct-2020. The MHRA also approved all protocol versions. The results of this trial will be disseminated through national and international presentations and peer-reviewed publications. Trial registration numberEudraCT Number: 2020-001684-89 ISRCTN Number: 40580903 Strengths and limitations of this trialO_LICATALYST will provide a rapid readout on the safety and proof-of-concept of candidate novel treatments C_LIO_LICATALYST will enable phase III trial resources to be focussed and allocated for agents with a high likelihood of success C_LIO_LICATALYST uses Bayesian multi-level models to allow for nesting of repeated measures data, with factors for each individual patient and treatment arm, and allowing for non-linear responses C_LIO_LICATALYST is not designed to provide a definitive signal on clinical outcomes C_LI

Vitamin D status and seroconversion for COVID-19 in UK healthcare workers who isolated for COVID-19 like symptoms during the 2020 pandemic. (preprint)

Background: It is clear that in UK healthcare workers, COVID-19 infections and deaths were more likely to be in staff who were of BAME origin. This has led to much speculation about the role of vitamin D in healthcare worker COVID-19 infections. We aimed to determine the prevalence of vitamin D deficiency in NHS staff who have isolated with symptoms suggestive of COVID-19 and relate this to vitamin D status. Methods: We recruited NHS healthcare workers between 12th to 22nd May 2020 as part of the COVID-19 convalescent immunity study (COCO). We measured anti-SARS-Cov-2 antibodies using a combined IgG, IgA and IgM ELISA (The Binding Site). Vitamin D status was determined by measurement of serum 25(OH)D3 using the AB SCIEX Triple Quad 4500 mass spectrometry system. Findings: Of the 392 NHS healthcare workers, 214 (55%) had seroconverted for COVID-19. A total of 61 (15{middle dot}6%) members of staff were vitamin D deficient (<30 nmol/l) with significantly more staff from BAME backgrounds or in a junior doctor role being deficient. Vitamin D levels were lower in those who were younger, had a higher BMI (>30 kg/m2), and were male. Multivariate analysis revealed that BAME and COVID-19 seroconversion were independent predictors of vitamin D deficiency. Staff who were vitamin D deficient were more likely to self-report symptoms of body aches and pains but importantly not the respiratory symptoms of cough and breathlessness. Vitamin D levels were lower in those COVID-19 positive staff who reported fever, but this did not reach statistical significance. Within the whole cohort there was an increase in seroconversion in staff with vitamin D deficiency compared to those without vitamin D deficiency (n=44/61, 72% vs n=170/331, 51%; p=0{middle dot} 003); this was particularly marked in the proportion of BAME males who were vitamin D deficient compared to non-vitamin D deficient BAME males (n=17/18, 94% vs n=12/23, 52%; p=0{middle dot}005). Multivariate analysis revealed that vitamin D deficiency was an independent risk factor for seroconversion (OR 2{middle dot}6, 95%CI 1{middle dot}41- 4{middle dot} 80; p=0{middle dot}002). Interpretation: In those healthcare workers who have isolated due to symptoms of COVID-19, those of BAME ethnicity are at the highest risk of vitamin D deficiency. Vitamin D deficiency is a risk factor for COVID-19 seroconversion for NHS healthcare workers especially in BAME male staff.

Characterization and Outcomes of ARDS secondary to pneumonia in patients with and without SARS-CoV-2: A single center experience (preprint)

ARDS is the major cause of mortality in patients with SARS-CoV-2 pneumonia. We report a single-centre study comparing the characteristics of ARDS patients with and without SARS-CoV-2. A greater proportion of SARS-CoV-2 patients were from an Asian ethnic group (p=0.002). SARS-CoV-2 patients had lower circulating leukocytes, neutrophils and monocytes (p<0.0001), but higher CRP (p=0.016) on ICU admission. SARS-CoV-2 patients required a longer duration of mechanical ventilation (p=0.01), but had lower vasopressor requirements (p=0.016). While the clinical syndromes of SARS-CoV-2 and CAP-ARDS are similar, the dysregulated inflammation observed in SARS-CoV-2 may contribute to the increased duration of respiratory failure.