Article Text

Download PDFPDF

Original research
Using a machine learning approach to predict mortality in critically ill influenza patients: a cross-sectional retrospective multicentre study in Taiwan
  1. Chien-An Hu1,
  2. Chia-Ming Chen2,
  3. Yen-Chun Fang3,
  4. Shinn-Jye Liang4,
  5. Hao-Chien Wang5,
  6. Wen-Feng Fang6,7,
  7. Chau-Chyun Sheu8,9,
  8. Wann-Cherng Perng10,
  9. Kuang-Yao Yang11,12,
  10. Kuo-Chin Kao13,14,
  11. Chieh-Liang Wu15,16,
  12. Chwei-Shyong Tsai3,
  13. Ming-Yen Lin1,
  14. Wen-Cheng Chao16,17
  15. On behalf of TSIRC (Taiwan Severe Influenza Research Consortium)
  1. 1Department of Information Engineering and Computer Science, Feng Chia University, Taichung, Taiwan
  2. 2Department of Applied Mathematics, National Chung Hsing University, Taichung, Taiwan
  3. 3Department of Management Information Systems, National Chung Hsing University, Taichung, Taiwan
  4. 4Division of Pulmonary and Critical Care, Department of Internal Medicine, China Medical University Hospital, Taichung, Taiwan
  5. 5Division of Chest Medicine, Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan
  6. 6Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, Kaohsiung Chang Gung Memorial Hospital, Kaohsiung, Taiwan
  7. 7Department of Respiratory Care, Chang Gung University of Science and Technology, Chiayi, Taiwan
  8. 8Division of Pulmonary and Critical Care Medicine, Kaohsiung Medical University Hospital, Kaohsiung, Taiwan
  9. 9School of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan
  10. 10Division of Pulmonary and Critical Care Medicine, Department of Internal Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, Taiwan
  11. 11Department of Chest Medicine, Taipei Veterans General Hospital, Taipei, Taiwan
  12. 12Institute of Emergency and Critical Care Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan
  13. 13Department of Thoracic Medicine, Chang Gung Memorial Hospital, Taoyuan, Taiwan
  14. 14Department of Respiratory Therapy, Chang Gung University College of Medicine, Taoyuan, Taiwan
  15. 15Center for Quality Management, Taichung Veterans General Hospital, Taichung, Taiwan
  16. 16Division of Chest, Department of Internal Medicine, Taichung Veterans General Hospital, Taichung, Taiwan
  17. 17Department of Critical Care Medicine, Taichung Veterans General Hospital, Taichung, Taiwan
  1. Correspondence to Dr Wen-Cheng Chao; cwc081{at}; Professor Ming-Yen Lin; linmy{at}


Objectives Current 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 design A cross-sectional retrospective multicentre study in Taiwan

Setting Eight medical centres in Taiwan.

Participants A 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 measures We 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.

Results The 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.

Conclusions We 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.

  • adult intensive & critical care
  • information technology
  • infectious diseases & infestations
  • adult intensive & critical care
  • thoracic medicine

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

View Full Text

Statistics from


  • C-AH and C-MC contributed equally.

  • Contributors Study concept and design: CA-H, CM-C, Y-CF, S-JL, H-CW, W-FF, C-CS, W-CP, K-YY, K-CK, C-LW, C-ST, M-YL and W-CC. Acquisition of data: Y-CF, CM-C and W-CC. Analysis and interpretation of data: W-CC, Y-CF, CM-C, CA-H and M-YL. Drafting the manuscript: W-CC.

  • Funding This study was supported in part by grants from Veterans General Hospitals and the University System of Taiwan Joint Research Program (VGHUST108-G2-4-2). The funders had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Taichung Veterans General Hospital CE16093A, National Taiwan University Hospital 201605036RIND, Taipei Veterans General Hospital 2016-05-020CC, Tri-Service General Hospital 1-105-05-086, Chang Gung Memorial Hospital 201600988B0, China Medical University Hospital 105-REC2-053(FR), Kaohsiung Medical University Hospital KUMHIRB-E(I)-20170097, Kaohsiung Chang Gung Memorial Hospital 201600988B0.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. All of the data and materials are provided in the manuscript and the supplemental data. The data set has been put in public Github, and is available via

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.