Yi HJ, et al. Ann Surg Treat Res. 2016 Mar;90(3):179-182. http://dx.doi.org/10.4174/astr.2016.90.3.179

Background: An artificial-intelligence (AI) model for predicting the prognosis or mortality of coronavirus disease 2019 (COVID-19) patients will allow efficient allocation of limited medical resources. We developed an early mortality prediction ensemble model for COVID-19 using AI models with initial chest X-ray and electronic health record (EHR) data. Results: We used convolutional neural network (CNN) models (Inception-ResNet-V2 and EfficientNet) for chest X-ray analysis and multilayer perceptron (MLP), Extreme Gradient Boosting (XGBoost), and random forest (RF) models for EHR data analysis. The Gradient-weighted Class Activation Mapping and Shapley Additive Explanations (SHAP) methods were used to determine the effects of these features on COVID-19. We developed an ensemble model (Area under the receiver operating characteristic curve of 0.8698) using a soft voting method with weight differences for CNN, XGBoost, MLP, and RF models. To resolve the data imbalance, we conducted F1-score optimization by adjusting the cutoff values to optimize the model performance (F1 score of 0.77). Conclusions: Our study is meaningful in that we developed an early mortality prediction model using only the initial chest X-ray and EHR data of COVID-19 patients. Early prediction of the clinical courses of patients is helpful for not only treatment but also bed management. Our results confirmed the performance improvement of the ensemble model achieved by combining AI models. Through the SHAP method, laboratory tests that indicate the factors affecting COVID-19 mortality were discovered, highlighting the importance of these tests in managing COVID-19 patients.

Hong KS, et al. Ann Surg Treat Res. 2014 Mar;86(3):152-160. https://doi.org/10.4174/astr.2014.86.3.152

Background: This study examined the experience of withholding or withdrawing life-sustaining treatment in patients hospitalized in the intensive care units (ICUs) of a tertiary care center. It also considers the role that intensivists play in the decision-making process regarding the withdrawal of life-sustaining treatment. Methods: We retrospectively analyzed the medical records of 227 patients who decided to withhold or withdraw life-sustaining treatment while hospitalized at Ewha Womans University Medical Center Mokdong between April 9 and December 31, 2018. Results: The 227 hospitalized patients included in the analysis withheld or withdrew from life-sustaining treatment. The department in which life-sustaining treatment was withheld or withdrawn most frequently was hemato-oncology (26.4%). Among these patients, the most common diagnosis was gastrointestinal tract cancer (29.1%). A majority of patients (64.3%) chose not to receive any life-sustaining treatment. Of the 80 patients in the ICU, intensivists participated in the decision to withhold or withdraw life-sustaining treatment in 34 cases. There were higher proportions of treatment withdrawal and ICU-to-ward transfers among the cases in whom intensivists participated in decision making compared to those cases in whom intensivists did not participate (50.0% vs. 4.3% and 52.9% vs. 19.6%, respectively). Conclusions: Through their participation in end-of-life discussions, intensivists can help patients' families to make decisions about withholding or withdrawing life-sustaining treatment and possibly avoiding futile treatments for these patients.

This study was designed to develop machine-learning models to predict COVID-19 mortality and identify its key features based on clinical characteristics and laboratory tests. For this, deep-learning (DL) and machine-learning (ML) models were developed using receiver operating characteristic (ROC) area under the curve (AUC) and F1 score optimization of 87 parameters. Of the two, the DL model exhibited better performance (AUC 0.8721, accuracy 0.84, and F1 score 0.76). However, we also blended DL with ML, and the ensemble model performed the best (AUC 0.8811, accuracy 0.85, and F1 score 0.77). The DL model is generally unable to extract feature importance; however, we succeeded by using the Shapley Additive exPlanations method for each model. This study demonstrated both the applicability of DL and ML models for classifying COVID-19 mortality using hospital-structured data and that the ensemble model had the best predictive ability.