Feature selection with ensemble learning for prostate cancer diagnosis from microarray gene expression.

Area of Science:

• Computational biology
• Bioinformatics
• Medical informatics

Background:

• Prostate cancer is a leading cause of death worldwide.
• Machine learning aids in cancer diagnosis, particularly for analyzing complex gene expression data from microarrays.
• Challenges remain in diagnosing prostate cancer from microarray data due to high dimensionality and limited sample sizes.

Purpose of the Study:

• To develop an effective machine learning method for prostate cancer detection using gene expression microarray data.
• To address the limitations of traditional methods in handling high-dimensional data with small sample sizes.
• To improve the accuracy of prostate cancer diagnosis through feature selection and ensemble learning.

Main Methods:

• Utilized Correlation Feature Selection (CFS) for identifying relevant genes.
• Employed Random Committee (RC) ensemble learning for improved diagnostic accuracy.
• Conducted experiments on a public benchmark dataset using a 10-fold cross-validation technique.

Main Results:

• The proposed CFS with RC ensemble learning achieved a high accuracy rate of 95.098%.
• This performance surpasses that of related methods applied to the same dataset.
• The approach effectively leverages microarray data for enhanced prostate cancer detection.

Conclusions:

• The combination of CFS and RC ensemble learning offers a powerful tool for accurate prostate cancer diagnosis.
• This method demonstrates significant potential in overcoming challenges associated with microarray data analysis in medical diagnosis.
• The findings highlight the utility of advanced machine learning techniques in improving cancer detection rates.