Search

Additional file 1: Includes frequency of patient censuses during the study period (Table S1); admission rates given patient census during rounding and daytime and during nighttime (Table S2); piecewise Poisson regression results (Table S3); sensitivity analysis with the different criterion of high mortality risk patients (Table S4); and sensitivity analysis including rare patient census occasions (Table S5).

Background: In COVID-19, high levels of granulocyte macrophage-colony stimulating factor (GM-CSF) and inflammatory myeloid cells correlate with disease severity, cytokine storm, and respiratory failure. With this rationale, we used lenzilumab, an anti-human GM-CSF monoclonal antibody, to treat patients with severe COVID-19 pneumonia. Methods: Hospitalized patients with COVID-19 pneumonia and risk factors for poor outcomes were treated with lenzilumab 600 mg intravenously for three doses through an emergency single-use IND application. Patient characteristics, clinical and laboratory outcomes, and adverse events were recorded. All patients receiving lenzilumab through May 1, 2020 were included in this report. Results: Twelve patients were treated with lenzilumab. Clinical improvement was observed in 11 out of 12 (92%), with a median time to discharge of 5 days. There was a significant improvement in oxygenation: The proportion of patients with SpO2/FiO2 < 315 at the end of observation was 8% vs. compared to 67% at baseline (p=0.00015). A significant improvement in mean CRP and IL-6 values on day 3 following lenzilumab administration was also observed (137.3 mg/L vs 51.2 mg/L, p = 0.040; 26.8 pg/mL vs 16.1 pg/mL, p = 0.035; respectively). Cytokine analysis showed a reduction in inflammatory myeloid cells two days after lenzilumab treatment. There were no treatment-emergent adverse events attributable to lenzilumab, and no mortality in this cohort of patients with severe COVID-19 pneumonia. Conclusions: In high-risk COVID-19 patients with severe pneumonia, GM-CSF neutralization with lenzilumab was safe and associated with improved clinical outcomes, oxygen requirement, and cytokine storm.

Understanding the temporal dynamics of COVID-19 patient phenotypes is necessary to derive fine-grained resolution of pathophysiology. Here we use state-of-the-art deep neural networks over an institution-wide machine intelligence platform for the augmented curation of 15.8 million clinical notes from 30,494 patients subjected to COVID-19 PCR diagnostic testing. By contrasting the Electronic Health Record (EHR)-derived clinical phenotypes of COVID-19-positive (COVIDpos, n=635) versus COVID-19-negative (COVIDneg, n=29,859) patients over each day of the week preceding the PCR testing date, we identify anosmia/dysgeusia (37.4-fold), myalgia/arthralgia (2.6-fold), diarrhea (2.2-fold), fever/chills (2.1-fold), respiratory difficulty (1.9-fold), and cough (1.8-fold) as significantly amplified in COVIDpos over COVIDneg patients. The specific combination of cough and diarrhea has a 3.2-fold amplification in COVIDpos patients during the week prior to PCR testing, and along with anosmia/dysgeusia, constitutes the earliest EHR-derived signature of COVID-19 (4-7 days prior to typical PCR testing date). This study introduces an Augmented Intelligence platform for the real-time synthesis of institutional knowledge captured in EHRs. The platform holds tremendous potential for scaling up curation throughput, with minimal need for retraining underlying neural networks, thus promising EHR-powered early diagnosis for a broad spectrum of diseases.