Discriminability of healthy versus cancer cells as a function of the number of genes considered.

<p>When measuring accuracy of cell classification as cancerous or healthy, one should consider both types of errors: false positives and false negatives (or more conventionally, true positives). This is illustrated by the Receiver Operating Characteristic (ROC) Curve. Lines indicate mean values, and error bars indicate bootstrapped 95% confidence intervals. Accuracy was measured using cross-validation; and chance value was determined using shuffle control.