Fig. 2

Flow diagram of data discovery process. Datasets were identified in online repositories, screened for eligibility, and included in further analysis. Datasets were excluded based on sample size, cancer progenitor, technology, or insufficient or no clinical features. Created with BioRender.com [14]