Comparison of Classifier Fusion Methods for Predicting Response to Anti HIV-1 Therapy
2008

Predicting Response to Anti HIV-1 Therapy

Sample size: 18467 publication 10 minutes Evidence: moderate

Author Information

Author(s): Altmann André, Rosen-Zvi Michal, Prosperi Mattia, Aharoni Ehud, Neuvirth Hani, Schülter Eugen, Büch Joachim, Struck Daniel, Peres Yardena, Incardona Francesca, Sönnerborg Anders, Kaiser Rolf, Zazzi Maurizio, Lengauer Thomas

Primary Institution: Max Planck Institute for Informatics

Hypothesis

Can different classifier fusion methods improve the prediction of response to combination antiretroviral therapy for HIV-1 patients?

Conclusion

The combined EuResist prediction engine is freely available and shows improved robustness in predicting therapy responses.

Supporting Evidence

  • The individual classifiers yielded similar performance.
  • Combination approaches performed equally well but did not significantly outperform the best individual classifier.
  • On smaller training set sizes, the combination significantly outperformed individual classifiers.

Takeaway

This study looks at how to better predict if HIV treatment will work by combining different prediction methods.

Methodology

The study compared various classifier fusion methods using data from the EuResist Integrated Database to predict therapy responses based on viral genotypes.

Potential Biases

The study noted that classifiers often agreed on incorrect labels, particularly in failing regimens.

Limitations

The performance of the combined methods did not significantly improve over the best individual classifier.

Participant Demographics

Patients included in the study were HIV-1 infected individuals from various countries, with data collected from multiple databases.

Statistical Information

P-Value

p<0.01

Confidence Interval

[30.6–36.1 million]

Statistical Significance

p<0.01

Digital Object Identifier (DOI)

10.1371/journal.pone.0003470

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication