Development and Validation of an Interpretable Machine Learning Model for the Prediction of Intubation in the Intensive Care Unit

Abstract

Background Since there is a limited ability to identify the need for intubation in the ICU, the objective of this study was to develop and validate an interpretable machine learning (ML) model to predict the need for intubation in ICU patients. Methods Seven widely used ML algorithms were applied to develop and validate prediction models. Adult patients from the Medical Information Mart for Intensive Care IV database who stayed in the ICU for longer than 24 hours were involved in developing the model. The model was then externally validated in the eICU-CRD database. In addition, the SHapley Additive exPlanations (SHAP) method was employed to illustrate the effects of the individual features attributed to the prediction model. Results A total of 11988 patients were included in the final cohort for this study. Among the seven constructed models, the CatBoost model achieved the best performance (AUC: 0.863). In the external validation, the efficacy of our model was also observed (AUC: 0.722). The Glasgow Coma Scale (GCS), body mass index (BMI), arterial partial pressure of oxygen (PaO2) and respiratory rate (RR) were the top 4 features of the CatBoost model with the greatest impact. Conclusions We have demonstrated the potential of machine-learning approaches for predicting the need for intubation in ICU patients. The SHAP method could improve the interpretability of machine-learning models and help clinicians better understand the reasoning behind the outcome. Trial registration: Not applicable.