Dataset Information

Predicting the microbial cause of community-acquired pneumonia: can physicians or a data-driven method differentiate viral from bacterial pneumonia at patient presentation?

ABSTRACT: BACKGROUND:Community-acquired pneumonia (CAP) requires urgent and specific antimicrobial therapy. However, the causal pathogen is typically unknown at the point when anti-infective therapeutics must be initiated. Physicians synthesize information from diverse data streams to make appropriate decisions. Artificial intelligence (AI) excels at finding complex relationships in large volumes of data. We aimed to evaluate the abilities of experienced physicians and AI to answer this question at patient admission: is it a viral or a bacterial pneumonia? METHODS:We included patients hospitalized for CAP and recorded all data available in the first 3-h period of care (clinical, biological and radiological information). For this proof-of-concept investigation, we decided to study only CAP caused by a singular and identified pathogen. We built a machine learning model prediction using all collected data. Finally, an independent validation set of samples was used to test the pathogen prediction performance of: (i) a panel of three experts and (ii) the AI algorithm. Both were blinded regarding the final microbial diagnosis. Positive likelihood ratio (LR) values >?10 and negative LR values

SUBMITTER: Lhommet C

PROVIDER: S-EPMC7060632 | biostudies-literature | 2020 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Predicting the microbial cause of community-acquired pneumonia: can physicians or a data-driven method differentiate viral from bacterial pneumonia at patient presentation?

Lhommet Claire C Garot Denis D Grammatico-Guillon Leslie L Jourdannaud Cassandra C Asfar Pierre P Faisy Christophe C Muller Grégoire G Barker Kimberly A KA Mercier Emmanuelle E Robert Sylvie S Lanotte Philippe P Goudeau Alain A Blasco Helene H Guillon Antoine A

BMC pulmonary medicine 20200306 1

<h4>Background</h4>Community-acquired pneumonia (CAP) requires urgent and specific antimicrobial therapy. However, the causal pathogen is typically unknown at the point when anti-infective therapeutics must be initiated. Physicians synthesize information from diverse data streams to make appropriate decisions. Artificial intelligence (AI) excels at finding complex relationships in large volumes of data. We aimed to evaluate the abilities of experienced physicians and AI to answer this question a ...[more]

PMID: 32143620

Similar Datasets

Project description:To develop and validate a model to predict resistance to community-acquired pneumonia antibiotics (CAP-resistance) among patients with healthcare-associated pneumonia (HCAP), and to compare the model's predictive performance to a model including only guideline-defined criteria for HCAP.Retrospective cohort study.Six Veterans Affairs Medical Centers in the northwestern United States.Culture-positive inpatients with HCAP.Patients were identified based upon guideline-defined criteria for HCAP. Relevant cultures obtained within 48 hours of admission were assessed to determine bacteriology and antibiotic susceptibility. Medical records for the year preceding admission were assessed to develop predictive models of CAP-resistance with logistic regression. The predictive performance of cohort-developed and guideline-defined models was compared.CAP-resistant organisms were identified in 118 of 375 culture-positive patients. Of guideline-defined criteria, CAP-resistance was associated (odds ratio (OR) [95% confidence interval (CI)]) with: admission from nursing home (2.6 [1.6-4.4]); recent antibiotic exposure (1.7 [1.0-2.8]); and prior hospitalization (1.6 [1.0-2.6]). In the cohort-developed model, CAP-resistance was associated with: admission from nursing home or recent nursing home discharge (2.3 [1.4-3.8]); positive methicillin-resistant Staphylococcus aureus (MRSA) history within 90 days of admission (6.4 [2.6-17.8]) or 91-365 days (2.3 [0.9-5.9]); cephalosporin exposure (1.8 [1.1-2.9]); recent infusion therapy (1.9 [1.0-3.5]); diabetes (1.7 [1.0-2.8]); and intensive care unit (ICU) admission (1.6 [1.0-2.6]). Area under the receiver operating characteristic curve (aROC [95% CI]) for the cohort-developed model (0.71 [0.65-0.77]) was significantly higher than for the guideline-defined model (0.63 [0.57-0.69]) (P = 0.01).Select guideline-defined criteria predicted CAP-resistance. A cohort-developed model based primarily on prior MRSA history, nursing home residence, and specific antibiotic exposures provided improved prediction of CAP-resistant organisms in HCAP.

Dataset Information

Predicting the microbial cause of community-acquired pneumonia: can physicians or a data-driven method differentiate viral from bacterial pneumonia at patient presentation?

Publications

Predicting the microbial cause of community-acquired pneumonia: can physicians or a data-driven method differentiate viral from bacterial pneumonia at patient presentation?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure