Additional file 1 of Machine learning-based identification of efficient and restrictive physiological subphenotypes in acute respiratory distress syndrome

Supplementary Material 1. Figure S1. Evaluation of the optimal number of clusters using Silhouette scores, Elbow method, and Akaike Information Criterion. The Silhouette method, Elbow method, and Akaike/Bayesian Information Criterionwere used to assess the optimal number of clusters. The best cluster numberis indicated by the red dashed line. Figure S2. Workflow for machine learning approaches. A detailed schematic illustrating the workflow from data preprocessing to the development of the final parsimonious linear model. Each step highlights key decisions and processes involved.