Computational protocol: Novel Algorithm for Non-Invasive Assessment of Fibrosis in NAFLD

Similar protocols

Protocol publication

[…] Several machine learning techniques were evaluated, namely logistic regression (logReg), k-nearest neighbors (knn), linear support vector machines (SVM), rule-based systems (RB), decision trees (DT) and random forest (RF). For the logistic regression, we used the implementation in the stats package of R (http://www.r-project.org) with standard settings. The k-nearest neighbor implementation in the R package class was used with k = 3. The SVM implementation in the package kernlab of R was used with the vanilladot kernel. For the rule-based systems we used the Part implementation provided in the R package RWeka. For the DTs we used the implementation in the rpart package and for the RFs we used the implementation in the randomForest package of R. In our application each RF consisted of 2000 randomly and independently grown decision trees. When using the trained RF for prediction, an unseen instance was assigned to the positive class voted for by at least 50% of the trees. The importance of each variable for the correct classification can be assessed by determining the decrease in Gini impurity . [...] All machine learning methods were validated using ten-fold leave-one-out cross-validation to assess for the different machine learning methods the mean prediction sensitivity, specificity, and accuracy (see formulas below) and the ability to generalize to unseen instances.For each test in the cross-validation, the sensitivity (SN), specificity (SP), and accuracy (AC) were calculated according to:with true positives TP, false positives FP, false negatives FN and true negatives TN. Furthermore, we calculated the Receiver Operating Characteristics (ROC) curve and the corresponding area under the curve (AUC) with ROCR . The ROC curve is built by plotting sensitivity vs. specificity for every possible cut-off between the two classes. […]

Pipeline specifications

Software tools randomforest, ROCR
Application Miscellaneous
Organisms Homo sapiens
Diseases Liver Diseases, Non-alcoholic Fatty Liver Disease