Using a machine learning approach to predict mortality in critically ill influenza patients: a cross-sectional retrospective multicentre study in Taiwan

• Published in: BMJ open Vol. 10; no. 2; p. e033898
• Main Authors: Hu, Chien-An, Chen, Chia-Ming, Fang, Yen-Chun, Liang, Shinn-Jye, Wang, Hao-Chien, Fang, Wen-Feng, Sheu, Chau-Chyun, Perng, Wann-Cherng, Yang, Kuang-Yao, Kao, Kuo-Chin, Wu, Chieh-Liang, Tsai, Chwei-Shyong, Lin, Ming-Yen, Chao, Wen-Cheng
• Format: Journal Article
• Language: English
• Published: England BMJ Publishing Group LTD 25.02.2020; BMJ Publishing Group
• Series: Original research
• Subjects: Accuracy; Adult; Critical Illness - mortality; Cross-Sectional Studies; Female; Hospital Mortality; Humans; Influenza; Influenza, Human - mortality; Intensive Care; Laboratories; Machine Learning; Male; Methods; Middle Aged; Models, Theoretical; Mortality; Outcome Assessment, Health Care; Patients; Pneumonia; Retrospective Studies; Sepsis; Taiwan; Variables; Ventilators; Taiwan; infectious diseases & infestations; thoracic medicine; information technology; adult intensive & critical care
• ISSN: 2044-6055; 2044-6055
• DOI: 10.1136/bmjopen-2019-033898

ObjectivesCurrent mortality prediction models used in the intensive care unit (ICU) have a limited role for specific diseases such as influenza, and we aimed to establish an explainable machine learning (ML) model for predicting mortality in critically ill influenza patients using a real-world severe influenza data set.Study designA cross-sectional retrospective multicentre study in TaiwanSettingEight medical centres in Taiwan.ParticipantsA total of 336 patients requiring ICU-admission for virology-proven influenza at eight hospitals during an influenza epidemic between October 2015 and March 2016.Primary and secondary outcome measuresWe employed extreme gradient boosting (XGBoost) to establish the prediction model, compared the performance with logistic regression (LR) and random forest (RF), demonstrated the feature importance categorised by clinical domains, and used SHapley Additive exPlanations (SHAP) for visualised interpretation.ResultsThe data set contained 76 features of the 336 patients with severe influenza. The severity was apparently high, as shown by the high Acute Physiology and Chronic Health Evaluation II score (22, 17 to 29) and pneumonia severity index score (118, 88 to 151). XGBoost model (area under the curve (AUC): 0.842; 95% CI 0.749 to 0.928) outperformed RF (AUC: 0.809; 95% CI 0.629 to 0.891) and LR (AUC: 0.701; 95% CI 0.573 to 0.825) for predicting 30-day mortality. To give clinicians an intuitive understanding of feature exploitation, we stratified features by the clinical domain. The cumulative feature importance in the fluid balance domain, ventilation domain, laboratory data domain, demographic and symptom domain, management domain and severity score domain was 0.253, 0.113, 0.177, 0.140, 0.152 and 0.165, respectively. We further used SHAP plots to illustrate associations between features and 30-day mortality in critically ill influenza patients.ConclusionsWe used a real-world data set and applied an ML approach, mainly XGBoost, to establish a practical and explainable mortality prediction model in critically ill influenza patients.