Novel Feature Selection Method for Cancer Gene Expression Data Achieves Excellent Performance

Strategic Insights -

Researchers at the University of Science and Technology Liaoning have developed a novel feature selection method, PF-PSS, to address the challenges of diagnosing cancer subtypes from high-dimensional and imbalanced cancer gene expression data. By employing a double-layer parallel embedded structure, PF-PSS achieves excellent performance in terms of classification accuracy and feature quantity selection. The method has been evaluated on 20 cancer gene expression datasets, demonstrating impressive results in various evaluation criteria and running time.

Key Takeaways:

  • The primary challenge in diagnosing cancer subtypes lies in the high dimensionality and sample imbalance of cancer gene expression data.
  • The novel feature selection method, PF-PSS, is structured into two distinct stages: a preliminary screening stage followed by a fine screening stage.
  • In the preliminary screening stage, a 'mean-median' method is employed to establish the threshold for the filter FS approach, while in the fine screening stage, the Lasso regression algorithm is used to refine the initial subset.
  • The PF-PSS method is evaluated on 20 cancer gene expression datasets, demonstrating excellent performance in terms of classification accuracy and feature quantity selection.
  • The proposed method achieved a classification accuracy of up to 100% for 9 of the datasets, and the proportion of selected features for 12 of the datasets remained within 7%.
  • The designed method has shown better performance in other evaluation criteria and running time.

Statistics:

  • The PF-PSS method achieved a classification accuracy of up to 100% for 9 out of 20 cancer gene expression datasets.
  • The proportion of selected features for 12 out of 20 datasets remained within 7%.
  • The minimum selected feature count reached 0.75% for one of the datasets.
  • The PF-PSS method demonstrated excellent performance in various evaluation criteria, including classification accuracy and feature quantity selection.
  • The method boasts better performance in running time compared to other evaluation methods.

Sources:

  • PF-PSS: a double-layer parallel embedded feature selection method for cancer gene expression data. Journal of Big Data, 2025,12(1):1-45.
  • (Journal of Big Data - https://journalofbigdata.springeropen.com)
  • University of Science and Technology Liaoning Researchers Report Recent Findings in Cancer (PF-PSS: a double-layer parallel embedded feature selection method for cancer gene expression data). Health & Medicine Week. June 20, 2025; p 5831.