Figure 5.

Serum biomarkers improve patient stratification in the study cohort. Logistic regression modelling was conducted on the cohort (n = 46) for classification as Healthy, BE or EAC, using patient risk factors of BMI, age and heartburn/reflux history alone (Clinical Risk Model), or risk factors plus serum biomarkers C9 and JAC-C9 (Biomarker Model). (a) Probability of health status was plotted as a violin plot against true classification. Boxes indicate the true health status, shift to the right indicates improved classification. (b) Receiver operating curve (ROC) for correctly predicting health status for Clinical Risk Model (red) vs. Biomarker Model (blue). See Table 2 for statistics.