Applying Gaussian Mixture Model for Clustering Analysis of Emergency Room Patients Based on Intubation Status

The study, conducted at two regional hospitals in Taichung, Taiwan, aimed to analyze emergency room patient data using Gaussian Mixture Model (GMM) for clustering based on intubation status. Out of 137,722 cases spanning January 1, 2017, to September 30, 2023, 1.14% underwent intubation. The study included the following variables: continuous variables such as WBC (White Blood Cell count), Hb (Hemoglobin), Hct (Hematocrit), MCV (Mean Corpuscular Volume), Blood Sugar, Creatinine levels, HR (Heart Rate), RR (Respiratory Rate), BT (Body Temperature), and SI (shock index). Additionally, categorical variables encompass Gender and Diabetes Mellitus (DM). Patients were divided into Rule In and Rule Out groups, with distinct intubation rates, 2.56% and 0.75%. Rule Out group, with a low intubation rate, identified patients with minimal intubation probability. We can infer that patients with elevated WBC, low Hb, low Hct, high blood sugar, high creatinine, high heart rate, and high shock index are more likely to require intubation compared to patients with normal values. Further research is needed to explore its application.