This study developed a predictive machine-learning model of in-hospital mortality based on pneumonia patient’s real-world clinical data. Predicting in-hospital mortality among high-risk patients is a critical issue in the medical field. Particularly, the prediction of risk occurrence in most pneumonia patients is based on pathological and radiological measurements, which are costly and time-consuming to analyze. Here, we propose a machine learning model that predicts the death of high-risk pneumonia patients based on actual clinical data, such as diagnosis codes, drug prescription codes, laboratory test codes, and nursing records from patients admitted with pneumonia. Specifically, we developed the model using data transformed into a 'Common Data Model (CDM)' format, which standardizes hospital data into a common framework. We demonstrate two types of predictive models: 1) traditional machine learning models and 2) stacking-based ensemble models. The mortality prediction model for patients diagnosed with pneumonia was evaluated in terms of precision, recall, F1-measure, and area under the curve (AUC). The experimental results show that our methods outperform standard classification models and, particularly, identify which factors based on the features input into the model influence the risk of death in patients. Since this model is based on CDM, it has the advantage of being extendable to other hospitals that hold this type of data in the future.

Borim Ryu received the Ph.D. degree in the Department of Biomedical engineering, college of medicine from Seoul National University, Seoul, Korea in 2022. She is currently an associate professor, in center for data science, biomedical research institute of Seoul National University Boramae medical center.  From 2015 to 2021, she was a Senior Researcher and Researcher at digital healthcare research center, in Seoul National University Bundang Hospital. In 2022, she worked as an assistant professor at Seoul National University Boramae Medical Center. Her research interests include AI for medicine, digital healthcare, international standards in biomedical domain.