Fig. 2From: Semantic categorization of Chinese eligibility criteria in clinical trials using machine learning methodsComparison of our proposed semantic categories with the previous works. The alignment of our 44 semantic categories (green) with Luo’s 27 semantic classes (red) and Van Spall’s 38 categories (blue). The histograms represent the prevalence. In Luo’s work, prevalence was calculated with 1578 randomly selected clinical trials from www.clinicaltrials.gov. In Van Spall’s work, prevalence was calculated with 283 random clinical trials published in high impact medical journals. In our study, prevalence was calculated with 272 HCC related clinical trials from ChiCTR. The lines represent the two categories have same meaning. The 13 semantic categories not linked with any other categories, are novel categories in Chinese eligibility criteriaBack to article page