Development and validation of a machine-learning model for prediction of hypoxemia after extubation in intensive care units

Abstract

Background: Extubation is the process of removing tracheal tubes so that patients maintain oxygenation while they start to breathe spontaneously. However, hypoxemia after extubation is an important issue for critical care doctors and is associated with patients' oxygenation, circulation, recovery, and incidence of postoperative complications. Accuracy and specificity of most related conventional models remain unsatisfactory. We conducted a predictive analysis based on a supervised machine-learning algorithm for the precise prediction of hypoxemia after extubation in intensive care units (ICUs).Methods: Data were extracted from the Medical Information Mart for Intensive Care (MIMIC)-IV database for patients over age 18 who underwent mechanical ventilation in the ICU. The primary outcome was hypoxemia after extubation, and it was defined as a partial pressure of oxygen <60 mmHg after extubation. Variables and individuals with missing values greater than 20% were excluded, and the remaining missing values were filled in using multiple imputation. The dataset was split into a training set (80%) and final test set (20%). All related clinical and laboratory variables were extracted, and logistics stepwise regression was performed to screen out the key features. Six different advanced machine-learning models, including logistics regression (LOG), random forest (RF), K-nearest neighbors (KNN), support-vector machine (SVM), eXtreme Gradient Boosting (XGBoost), and Light Gradient Boosting Machine (LightGBM), were introduced for modelling. The best performance model in the first cross-validated dataset was further fine-tuned, and the final performance was assessed using the final test set.Results: A total of 14,777 patients were included in the study, and 1,864 of the patients' experienced hypoxemia after extubation. After training, the RF and LightGBM models were the strongest initial performers, and the area under the curve (AUC) using RF was 0.780 [95% confidence interval (CI), 0.755-0.805] and using LightGBM was 0.779 (95% CI, 0.752-0.806). The final AUC using RF was 0.792 (95% CI, 0.771-0.814) and using LightGBM was 0.792 (95% CI, 0.770-0.815).Conclusions: Our machine learning models have considerable potential for predicting hypoxemia after extubation, which help to reduce ICU morbidity and mortality.

Predicting Hypoxia Using Machine Learning: Systematic Review

“…Most studies (9/12, 75%) analyzed a large sample size of 500 or more patients [ 15 , 16 , 18 - 22 , 25 , 26 ]. Data from the publicly available databases Medical Information Mart for Intensive Care and eICU Collaborative Research Database were used in 4 of the studies [ 15 , 16 , 21 , 22 ], whereas 3 studies relied on data collected via an anesthesia information management system (AIMS) [ 16 , 18 , 20 ].…”

“…Most studies (9/12, 75%) analyzed a large sample size of 500 or more patients [ 15 , 16 , 18 - 22 , 25 , 26 ]. Data from the publicly available databases Medical Information Mart for Intensive Care and eICU Collaborative Research Database were used in 4 of the studies [ 15 , 16 , 21 , 22 ], whereas 3 studies relied on data collected via an anesthesia information management system (AIMS) [ 16 , 18 , 20 ]. AIMSs are widely adopted hardware and software solutions that are integrated into a hospital’s electronic health record system and are used to manage and document a patient’s perioperative measurements [ 27 , 28 ].…”

“…Figure 2 [ 15 - 26 ] gives an overview of the models and the number of variables used in each study. Exclusively conventional ML algorithms were applied in 5 of the identified studies [ 16 , 17 , 20 , 22 , 23 ], whereas 7 studies included deep learning algorithms [ 15 , 18 , 19 , 21 , 24 - 26 ]. Models based on logistic regression were used most often (n=4) [ 18 , 21 - 23 ], followed by artificial neural networks (n=3) [ 21 , 24 , 25 ].…”

“…The remaining 5 studies relied on at least 6 variables [ 16 , 19 , 20 , 22 , 25 ]. The most frequently used variable sources were oximetry measurements (9/12, 75%) [ 15 - 22 , 25 ] and static patient characteristics such as age (5/12, 42%) [ 16 , 19 , 20 , 23 , 25 ]. Additionally, a single study relied on diagnostic images of the chest to make predictions [ 26 ].…”

“…Overview Table 1 presents the characteristics of all included studies and gives an overview of the best-performing model in each study, divided into conventional ML and deep learning models for studies including both. The studies were conducted in the United States [15][16][17][18][19][20][21][22], China [23,24], Germany [25], and the United Arab Emirates [26]. Half (6/12, 50%) of them were published after 2020 [15,16,19,21,22,26].…”