Machine Learning Accurately Predicts Need for Critical Care Support in Patients Admitted to Hospital for Community-Acquired Pneumonia.

Karen C Tran, Josh Douglas, Jennifer Yoon, Francois Lellouche, Juthaporn Cowan, Anne McCarthy, George S Chen, John C Marshall, Joel Singer, Terry Lee, Alexis F Turgeon, Alexandra Binnie, Francois Lamontagne, Donald C Vinh, Jennifer L Y Tsang, Peter Daley, Patrick Archambault, Allison McGeer, James A Russell, Robert Fowler, John H Boyd, Brent W Winston, Keith R Walley, David M Maslove, Todd C Lee, Matthew P Cheng

Journal: Critical care explorations 2025;7(6):e1262

PMID: 40443788

Free Full Text on PubMed Central View on PubMed

Abstract

OBJECTIVES

Hospitalized community-acquired pneumonia (CAP) patients are admitted for ventilation, vasopressors, and renal replacement therapy (RRT). This study aimed to develop a machine learning (ML) model that predicts the need for such interventions and compare its accuracy to that of logistic regression (LR).

DESIGN

This retrospective observational study trained separate models using random-forest classifier (RFC), support vector machines (SVMs), Extreme Gradient Boosting (XGBoost), and multilayer perceptron (MLP) to predict three endpoints: eventual use of invasive ventilation, vasopressors, and RRT during hospitalization. RFC-based models were overall most accurate in a derivation COVID-19 CAP cohort and were validated in one COVID-19 CAP and two non-COVID-19 CAP cohorts.

SETTING

This study is part of the Community-Acquired Pneumonia: Toward InnoVAtive Treatment (CAPTIVATE) Research program.

PATIENTS

Two thousand four hundred twenty COVID-19 and 1909 non-COVID-19 CAP patients over 18 years old hospitalized and not needing invasive ventilation, vasopressors, and RRT on the day of admission were included.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Performance was evaluated with area under the receiver operating characteristic curve (AUROC) and accuracy. RFCs performed better than XGBoost, SVM, and MLP models. For comparison, we evaluated LR models in the same cohorts. AUROC was very high ranging from 0.74 to 0.95 in predicting ventilation, vasopressors, and RRT use in our derivation and validation cohorts. ML used and variables such as Fio, Glasgow Coma Scale, and mean arterial pressure to predict ventilator, vasopressor use, creatinine, and potassium to predict RRT use. LR was less accurate than ML, with AUROC ranging 0.66 to 0.8.

CONCLUSIONS

A ML algorithm more accurately predicts need of invasive ventilation, vasopressors, or RRT in hospitalized non-COVID-19 CAP and COVID-19 patients than regression models and could augment clinician judgment for triage and care of hospitalized CAP patients.

Address: University of British Columbia, Vancouver, BC, Canada.; Centre for Advancing Health Outcomes, St. Paul's Hospital, University of British Columbia, Vancouver, BC, Canada.; Critical Care Medicine, Niagara Health Knowledge Institute, St Catharines, ON, Canada.; Critical Care Medicine, McMaster University, Hamilton, ON, Canada.; Critical Care Department, William Osler Health System, Brampton, ON, Canada.; Critical Care Medicine, Algarve Biomedical Centre, Faro, Portugal.; Critical Care Medicine, Centro Hospitalar Universitário do Algarve, Faro, Portugal.; Infectious Disease, Ottawa Research Institute, University of Ottawa, Ottawa, ON, Canada.; St. George Hospital, Levis, QC, Canada.; CHU de Québec-Université Laval Research Center, Population Health and Optimal Health Practices Unit, Trauma- Emergency- Critical Care Medicine, Québec City, QC, Canada.; Department of Anesthesiology and Critical Care Medicine, Division of Critical Care Medicine, Faculty of Medicine, Université Laval, Québec City, QC, Canada.; Critical Care Medicine, Humber River Hospital, Toronto, ON, Canada.; Critical Care Medicine, University of Sherbrooke, Sherbrooke, QC, Canada.; Mt. Sinai Hospital, University of Toronto, Toronto, ON, Canada.; Critical Care Medicine, Lion's Gate Hospital, North Vancouver, BC, Canada.; Infectious Disease, Memorial University of Newfoundland, St. John's, NL, Canada.; Critical Care Medicine, Sunnybrook Health Sciences Centre, Toronto, ON, Canada.; Department of Critical Care, Kingston General Hospital and Queen's University, Kingston, ON, Canada.; Departments of Critical Care Medicine, Medicine and Biochemistry and Molecular Biology, Foothills Medical Centre, University of Calgary, Calgary, AB, Canada.; Division of Infectious Disease, McGill University, Montreal, QC, Canada.; Division of General Internal Medicine, Vancouver General Hospital, Vancouver, BC, Canada.; Centre for Heart Lung Innovation, St. Paul's Hospital, University of British Columbia, Vancouver, BC, Canada.; Division of Critical Care Medicine, St. Paul's Hospital, University of British Columbia, Vancouver, BC, Canada.; Department of Surgery, St. Michael's Hospital, Toronto, ON, Canada.

DOI: 10.1097/CCE.0000000000001262

Link outs

Free resources

Full Text Sources:

Medical:

https://medlineplus.gov/covid19coronavirusdisease2...

Miscellaneous:

https://assays.cancer.gov/CPTAC-325

Subscription / membership required

Full Text Sources:

http://ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=linko...

MeSH Terms: Humans, Retrospective Studies, Community-Acquired Infections, Machine Learning, Male, Female, COVID-19, Middle Aged, Aged, Critical Care, Hospitalization, Respiration, Artificial, Renal Replacement Therapy, Pneumonia, Vasoconstrictor Agents, SARS-CoV-2, Community-Acquired Pneumonia