Fig. 4

The distribution of common prevalence scores between IPV and TBI cohorts, as compared to the distribution of common prevalence scores between the IPV cohort and each of the control cohorts. For each pair of cohorts, the respective histogram shows the distribution of common prevalence scores across all terms that have non-zero frequency in both cohorts. Common prevalence for a term with respect to two cohorts is defined as the smaller of the two log-odds-ratios of the frequency of the term in each of the two cohorts (thus a term has higher common prevalence if it has a higher log-odds-ration on both cohorts). On both panels, the cyan histogram shows the distribution of common prevalence scores for IPV, in comparison to a IPV and each of the acute condition cohorts, b IPV and each of the accident cohorts. The dashed and solid lines on the left panel show the cut-off lines for a false discovery rate of respectively 5% and 1% based on the acute condition controls, as indicated by the color of the line