Predicting Survival Outcomes with Gene Subsets
Author Information
Author(s): Matsui Shigeyuki
Primary Institution: Kyoto University
Hypothesis
Can subsets of significant genes improve prognostic predictions in cancer studies using microarrays?
Conclusion
The developed methodology can enhance prognostic predictions by using a subset of significant genes.
Supporting Evidence
- The methodology was evaluated using both simulated and real data.
- The predictive capability of the selected genes was substantial when combined with standard prognostic factors.
- The study demonstrated that gene expression data can provide additional prognostic information independent of established factors.
Takeaway
This study shows how we can use important genes to better predict how long patients might live after treatment.
Methodology
The study developed a methodology using Cox proportional hazard models and cross-validated log partial likelihood to predict survival outcomes.
Potential Biases
Selection bias may occur if gene filtering is not re-performed for each cross-validation round.
Limitations
The methodology's predictive accuracy can be difficult to interpret for non-statisticians.
Participant Demographics
The study involved 240 patients with diffuse large B-cell lymphoma.
Statistical Information
P-Value
0.0076
Statistical Significance
p<0.0005
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website