Final random forest 17 gene/probe model to predict prostate cancer systemic progression after a rising PSA following radical prostatectomy

*Mean Gini Decrease for a variable is the average (over all random forest trees) decrease in node impurities from recursive partitioning splits on that variable. For classification, the node impurity is measured by the Gini index. The Gini index is the weighted average of the impurity in each branch, with impurity being the proportion of incorrectly classified samples in that branch. The larger the Gini decrease, the fewer the misclassification impurities.**Genes mapped to 8q24***Single probes for C8orf53 and CDK10 were selected. The Mean Gini Decrease for these probes are derived from an independent random forest analysis of the all probes separately.