A generalizable 29-mRNA neural-network classifier for acute bacterial and viral infections

Abstract

Improved identification of bacterial and viral infections would reduce morbidity from sepsis, reduce antibiotic overuse, and lower healthcare costs. Here, we develop a generalizable hostgene-expression-based classifier for acute bacterial and viral infections. We use training data (N = 1069) from 18 retrospective transcriptomic studies. Using only 29 preselected host mRNAs, we train a neural-network classifier with a bacterial-vs-other area under the receiver-operating characteristic curve (AUROC) 0.92 (95% CI 0.90-0.93) and a viral-vsother AUROC 0.92 (95% CI 0.90-0.93). We then apply this classifier, inflammatix-bacterialviral-noninfected-version 1 (IMX-BVN-1), without retraining, to an independent cohort (N = 163). In this cohort, IMX-BVN-1 AUROCs are: bacterial-vs.-other 0.86 (95% CI 0.77-0.93), and viral-vs.-other 0.85 (95% CI 0.76-0.93). In patients enrolled within 36 h of hospital admission (N = 70), IMX-BVN-1 AUROCs are: bacterial-vs.-other 0.92 (95% CI 0.83-0.99), and viral-vs.-other 0.91 (95% CI 0.82-0.98). With further study, IMX-BVN-1 could provide a tool for assessing patients with suspected infection and sepsis at hospital admission.

Cytokine profile in plasma of severe COVID-19 does not differ from ARDS and sepsis

“…Briefly, the Stanford ICU Biobank recruits patients at risk for development of respiratory failure and ARDS admitted to Stanford Hospital as previously described. 11 Subjects are eligible for enrollment when decision to admit to ICU is made, either from the Emergency Department or the hospital wards, with goal enrollment in <24h of ICU transfer. All 28 patients were phenotyped for ARDS and sepsis by 2-physician consensus (AJR and JEL), based on the Berlin Criteria and Sepsis-2 criteria and using all available hospital clinical data including history, physical exam, laboratory and microbiologic data, invasive monitoring data, autopsy results, and physician summaries.…”

Cytokine profile in plasma of severe COVID-19 does not differ from ARDS and sepsis

“…Twelve critically ill subjects with ARDS and 16 patients with sepsis who had been enrolled in the Stanford ICU biobank between 2015 and 2018 were selected for comparison. Briefly, the Stanford ICU Biobank recruits patients at risk for development of respiratory failure and ARDS admitted to Stanford Hospital as previously described ( 15 ). Subjects are eligible for enrollment when a decision to admit to ICU is made, either from the Emergency Department or the hospital wards, with goal enrollment in < 24 hours of ICU transfer.…”

SIMON: open-source knowledge discovery platform

“…ML has a profound impact on biological research [10][11][12] , including genomics 13 , proteomics [14][15][16] , cell image analysis 17 , drug discovery and development 18 , and cell phenotyping 6,19,20 which revolutionized our understanding of biological complexity. Recently, using systems-level analysis of genetic, transcriptional, and proteomic signatures to predict patients' response to vaccines 21,22 , therapies and disease progression [23][24][25][26][27] , ML has become primary computational approach used in the 'precision medicine' 28 .…”