Media.Player

A novel statistical test and software for assessing the significance of difference in survivorship among survival classes from prognostic models

Detailed Technology Description

None

*Abstract

The subject invention is a novel statistical test (F* Test) and associated software (written in Perl) to detect significant differences in the survivorship of classes of patients (Predictive Survivorship Groups). Traditional survivorship analysis methods include the Log Rank Test, Kaplan-Meier Curves and Cox Proportional Hazard Models. The F* Test is a variation of the standard F Test. The F* Test takes into account the error rate of the model by studying the variance of the predicted survivorship in each predicted group around an unbiased mean group survivorship.Advantages1) Tracks accuracy of the prediction model with higher parity than the log-rank test.2) Avoids wasted research dollars in clinical trials of ineffective prognostic models.3) Protects patients from models that do not provide any real benefit for therapy selection.Stage of DevelopmentPrototype software written and tested.

*Principal Investigator

Name: Holly Berty

Department: Med-Biomedical Informatics

Name: Shi Haiwen

Department: PCI Basic Research

Name: James Lyons-Weiler, Assistant Professor

Department: Health Sciences Core Research Facilities

Country/Region

USA