Fig. 5

Machine learning based liver cirrhosis detection. A ROC curve displaying the AUC for the logistic regression model for recurrency threshold of fifteen. B Confusion matrix showing the percentage of samples that were correctly and incorrectly classified as liver cirrhosis patients or healthy controls, for recurrency threshold of fifteen. C Logistic regression classification coefficients. D ROC curve displaying the AUC for the XGBoost classification model, for recurrency threshold of fifteen. E AUC score relative to number of top frequentmers used for logistic regression. Gray lines display the confidence intervals from the ten folds. The blue line shows the mean AUC score across the ten folds