Semi-supervised methods to predict patient survival from gene expression data.

摘要: An important goal of DNA microarray research is to develop tools diagnose cancer more accurately based on the genetic profile a tumor. There are several existing techniques in literature for performing this type diagnosis. Unfortunately, most these assume that different subtypes already known exist. Their utility limited when such have not been previously identified. Although methods identifying exist, do work well all datasets. It would be desirable procedure find applicable wide variety circumstances. Even if no information about possible certain form cancer, clinical patients, as their survival time, often available. In study, we some procedures utilize both gene expression data and identify use knowledge future patients. These were successfully applied publicly available We present diagnostic predict patients times previous This has potential powerful tool diagnosing treating cancer.