Machine Learning Model Utilization for Mortality Prediction in Mechanically Ventilated ICU Patients
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
The requirement of mechanical ventilation has increased in recent years. Patients in the intensive care unit (ICU) who undergo mechanical ventilation often experience serious illness and critical physiological states, contributing to a high risk of mortality. Prediction of mortality for mechanically ventilated ICU patients offers an early alert mechanism to help physicians implement more targeted treatments to mitigate the risk of mortality.
Methods
We extracted medical information of patients with invasive mechanical ventilation in ICU admission from the Medical Information Mart for Intensive Care III (MIMIC III) dataset. This includes demographics, disease severity, diagnosis, and laboratory test results. Patients who met inclusion criteria were randomly divided into the train set (n=11,549, 70%), the test set (n=2,475, 15%), and the validation set (n=2,475, 15%). The oversampling method was utilized to resolve the imbalance dataset. After literature research, clinical expertise and ablation study, we selected 12 variables for model establishment, which were less than the 66 features in the best existing literature. We proposed a deep learning neural network model to predict the ICU mortality of mechanically ventilated patients, and established 7 baseline machine learning (ML) models for comparison, including k-nearest neighbors (KNN), logistic regression, decision tree, random forest, bagging, XGBoost, and support vector machine (SVM). Area under the Receiver Operating Characteristic Curve (AUROC) was used as an evaluation metric for model performance.
Results
By using 16,499 mechanically ventilated patients in the MIMIC III database, the deep learning model demonstrated the best performance, showing a 7.06% improvement over the best existing literature. The neural network achieved an AUROC score of 0.879 with 95% Confidence Interval (CI) [0.861-0.896] and an accuracy score of 0.859 on the test set. These results far exceeded those of the best existing literature.
Conclusions
The proposed model demonstrated an exceptional ability to predict ICU mortality among mechanically ventilated patients. Through SHAP analysis, we found that respiratory failure is a significant indicator of the mortality prediction compared to other related respiratory dysfunction diseases. We also incorporated mechanical ventilation duration as a variable for the first time in our prediction model. We observed that patients with higher mortality rates tended to have longer mechanical ventilation times. This highlights the model’s potential in guiding clinical decisions by indicating that longer mechanical ventilation may not necessarily enhance patient survival chances.
