Related Experiment Video
Updated: May 20, 2026
07:35
Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances
Published on: October 11, 2018
A Partitioning Based Adaptive Method for Robust Removal of Irrelevant Features from High-dimensional Biomedical
Guodong Liu1, Lan Kong, Vanathi Gopalakrishnan
1Pennsylvania State University, Hershey, PA; University of Pittsburgh, Pittsburgh, PA.
Summary
A new method, Partitioning based Adaptive Irrelevant Feature Eliminator (PAIFE), effectively reduces dimensionality in biomedical data by identifying conditional feature relevancies. PAIFE outperforms existing methods, improving classification accuracy on high-dimensional clinical datasets.
Area of Science:
- Bioinformatics
- Computational Biology
- Data Science
Background:
- High-dimensional biomedical datasets pose significant challenges for analysis and model building.
- Existing dimensionality reduction techniques may struggle to identify features with conditional relevancy.
Purpose of the Study:
- To introduce a novel dimensionality reduction method, Partitioning based Adaptive Irrelevant Feature Eliminator (PAIFE).
- To enhance the identification of relevant features and removal of irrelevant ones in high-dimensional biomedical data.
Main Methods:
- PAIFE evaluates feature-target relationships across the entire dataset and its subsets.
- It adaptively selects appropriate feature evaluation strategies, statistical tests, and parameters.
- The method is designed as a third-party data pre-processing tool.
Main Results:
- PAIFE outperformed state-of-the-art methods in removing irrelevant features while retaining relevant ones on synthetic datasets.
- Significant irrelevant feature reduction was observed in real-world genomic and proteomic datasets.
- Classification models using PAIFE-selected features achieved comparable or improved performance.
Conclusions:
- PAIFE is an effective tool for dimensionality reduction in high-dimensional biomedical data.
- The method successfully handles features with conditional relevancies.
- PAIFE can improve the efficiency and performance of downstream machine learning tasks in clinical settings.