Machine learning predicts mortality based on analysis of ventilation parameters of critically ill patients: multi-centre validation

BMC Medical Informatics and Decision Making

Pravin Amin¹  Amine Ali Zeggwagh²  Richard Rezar³  Bernhard Wernly³  Alfonso Muriel⁴  Marco González⁵  Andrew Bersten⁶  Gee Young Suh⁷  Raphael Romano Bruno⁸  Christian Jung⁸  Malte Kelm⁸  Yuda Sutherasan⁹  Behrooz Mamandipoor^1,10  Venet Osmani^1,10  Nicolas Nin^1,11  Fekri Abroug^1,12  Fernando Ríos^1,13  Maria del Carmen Marín^1,14  Marco Antonio Soares^1,15  Andrés Esteban^1,16  Fernando Frutos-Vivar^1,16  Oscar Peñuelas^1,16  Manuel Jibaja^1,17  Lorenzo del-Sorbo^1,18  Nahit Cakar^1,19  Konstantinos Raymondos^2,20  Dimitros Matamis^2,21  Bin Du^2,22  Bruno Valle Pinheiro^2,23  Antonio Anzueto^2,24  Arnaud W. Thille^2,25  Salvatore M. Maggiore^2,26 

[1] Bombay Hospital Institute of Medical Sciences, Mumbai, India;Centre Hospitalier Universitarie Ibn Sina - Mohammed V University, Rabat, Morocco;Clinic of Internal Medicine II, Department of Cardiology, Paracelsus Medical University of Salzburg, 5020, Salzburg, Austria;Clinic of Internal Medicine II, Department of Cardiology, Paracelsus Medical University of Salzburg, 5020, Salzburg, Austria;Unidad de Bioestadística Clinica Hospital Ramón y Cajal, Instituto Ramón y Cajal de Investigaciones Sanitarias (IRYCIS) & Centro de Investigación en Red de Epidemiología y Salud Pública (CIBERESP), Madrid, Spain;Clínica Medellín & Universidad Pontificia Bolivariana, Medellín, Colombia;Department of Critical Care Medicine, Flinders University, Adelaide, South Australia, Australia;Department of Critical Care Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea;Division of Cardiology, Pulmonology and Vascular Medicine, Medical Faculty, University of Düsseldorf, Moorenstraße 5, 40225, Düsseldorf, Germany;Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand;Fondazione Bruno Kessler Research Institute, Trento, Italy;Hospital Español, Montevideo, Uruguay;Hospital Fattouma Bourguina, Monastir, Tunisia;Hospital Nacional Alejandro Posadas, Buenos Aires, Argentina;Hospital Regional 1° de Octubre, Instituto de Seguridad Y Servicios Sociales de Los Trabajadores del Estado (ISSSTE), México, DF, México;Hospital Universitario Sao Jose, Belo Horizonte, Brazil;Hospital Universitario de Getafe & Centro de Investigación en Red de Enfermedades Respiratorias (CIBERES), Madrid, Spain;Hospital de Especialidades Eugenio Espejo, Quito, Ecuador;Interdepartmental Division of Critical Care Medicine, Toronto, ON, Canada;Istanbul Faculty of Medicine, Istanbul, Turkey;Medizinische Hochschule Hannover, Hannover, Germany;Papageorgiou Hospital, Thessaloniki, Greece;Peking Union Medical College Hospital, Beijing, People’s Republic of China;Pulmonary Research Laboratory, Federal University of Juiz de Fora, Juiz de Fora, Brazil;South Texas Veterans Health Care System and University of Texas Health Science Center, San Antonio, TX, USA;University Hospital of Poitiers, Poitiers, France;Università Degli Studi G. d’Annunzio Chieti e Pescara, Chieti, Italy;

关键词: Critical care medicine;    Machine learning;    ICU;    Risk stratification;    Mechanical ventilation;   

DOI  :  10.1186/s12911-021-01506-w

来源: Springer

【 摘 要 】

BackgroundMechanical Ventilation (MV) is a complex and central treatment process in the care of critically ill patients. It influences acid–base balance and can also cause prognostically relevant biotrauma by generating forces and liberating reactive oxygen species, negatively affecting outcomes. In this work we evaluate the use of a Recurrent Neural Network (RNN) modelling to predict outcomes of mechanically ventilated patients, using standard mechanical ventilation parameters.MethodsWe performed our analysis on VENTILA dataset, an observational, prospective, international, multi-centre study, performed to investigate the effect of baseline characteristics and management changes over time on the all-cause mortality rate in mechanically ventilated patients in ICU. Our cohort includes 12,596 adult patients older than 18, associated with 12,755 distinct admissions in ICUs across 37 countries and receiving invasive and non-invasive mechanical ventilation. We carry out four different analysis. Initially we select typical mechanical ventilation parameters and evaluate the machine learning model on both, the overall cohort and a subgroup of patients admitted with respiratory disorders. Furthermore, we carry out sensitivity analysis to evaluate whether inclusion of variables related to the function of other organs, improve the predictive performance of the model for both the overall cohort as well as the subgroup of patients with respiratory disorders.ResultsPredictive performance of RNN-based model was higher with Area Under the Receiver Operating Characteristic (ROC) Curve (AUC) of 0.72 (± 0.01) and Average Precision (AP) of 0.57 (± 0.01) in comparison to RF and LR for the overall patient dataset. Higher predictive performance was recorded in the subgroup of patients admitted with respiratory disorders with AUC of 0.75 (± 0.02) and AP of 0.65 (± 0.03). Inclusion of function of other organs further improved the performance to AUC of 0.79 (± 0.01) and AP 0.68 (± 0.02) for the overall patient dataset and AUC of 0.79 (± 0.01) and AP 0.72 (± 0.02) for the subgroup with respiratory disorders.ConclusionThe RNN-based model demonstrated better performance than RF and LR in patients in mechanical ventilation and its subgroup admitted with respiratory disorders. Clinical studies are needed to evaluate whether it impacts decision-making and patient outcomes.Trial registration: NCT02731898 (https://clinicaltrials.gov/ct2/show/NCT02731898), prospectively registered on April 8, 2016.

【 授权许可】

CC BY