Skip to main content
. 2023 May 9;13:7544. doi: 10.1038/s41598-023-34303-8

Figure 1.

Figure 1

Study design. The first stage was to establish the cohort. Our pipeline for creating models for PE risk stratification consisted of collecting two data modalities: clinical data (EHR) including patient demographics, comorbidities, vital signs, and relevant labs, and CTPA scans. We risk-stratified the diagnosed PE using the 30-day all-cause mortality label from the imaging data and the clinical variable data independently. Finally, these data were linked using the unimodal classifiers from the previous stage to develop, analyze, and benchmark the models for PE prognosis.

(Icon images from Microsoft PowerPoint 2023).