Skip to main content
. 2019 Aug 8;14(8):e0219728. doi: 10.1371/journal.pone.0219728

Fig 3. Performance of the predictors on an independent cohort.

Fig 3

A) Performance of our models and other published models in prediction of primary CV outcome for patients from the ACCORD trial. All models were trained on the SPRINT data. For our models, mean and standard errors of 50 repetitions of the internal sampling are shown (see S1 File). As in Fig 1, sample sizes vary between the baseline and longitudinal models due to excluded samples. B) Hazard ratios for different risk groups of ACCORD patients. All ACCORD patients were divided into three equally sized risk groups and HR’s were calculated between the groups as explained for Fig 1B. HR between the high and low risk groups: 6.010 (3.110–7.763 95% CI); between high and medium: 2.573 (2.129–3.110 95% CI); between medium and low risk groups: 2.384 (1.801–3.157 95% CI). All HR P-values < 5.5*10^-9.