Flexible boosting of accelerated failure time models
2008

New Boosting Algorithm for Survival Models

Sample size: 50 publication 10 minutes Evidence: high

Author Information

Author(s): Matthias Schmid, Torsten Hothorn

Primary Institution: Friedrich-Alexander-Universität Erlangen-Nürnberg

Hypothesis

Can a new boosting algorithm improve the fitting of parametric accelerated failure time models for survival data?

Conclusion

The new boosting algorithm outperforms traditional methods when the proportional hazards assumption is questionable.

Supporting Evidence

  • The new algorithm was shown to outperform traditional boosting methods in cases where the proportional hazards assumption is violated.
  • Simulations indicated that the new boosting algorithm closely approximates maximum likelihood estimates in low-dimensional settings.
  • The analysis of microarray data demonstrated the algorithm's effectiveness in predicting survival outcomes.

Takeaway

This study created a new way to predict how long patients might live based on their gene data, and it works better than older methods when the usual rules don't apply.

Methodology

The study developed a boosting algorithm that estimates both the prediction function and a scale parameter simultaneously using negative log likelihood as a loss function.

Potential Biases

Potential biases may arise from the selection of genes and the specific modeling assumptions made.

Limitations

The study primarily focused on a specific dataset and may not generalize to all types of survival data.

Participant Demographics

The study involved 50 patients with stage II colon adenocarcinoma.

Statistical Information

P-Value

<0.001

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.1186/1471-2105-9-269

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication