A framework model using multifilter feature selection to enhance colon cancer classification.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Gene expression profiling is vital for diagnosing diseases like cancer.
• Accurate biomarker gene selection is crucial for effective cancer detection and diagnosis.
• Colon cancer is a prevalent disease requiring rapid and precise diagnostic methods.

Purpose of the Study:

• To develop and evaluate a two-stage multifilter hybrid feature selection model for improved colon cancer classification.
• To enhance the diagnostic process and accelerate drug discovery for colon cancer.

Main Methods:

• A two-stage hybrid feature selection framework combining Information Gain and Genetic Algorithm.
• Gene filtering and ranking using the minimum Redundancy Maximum Relevance (mRMR) technique.
• Analysis of selected genes using machine learning algorithms such as Decision Tree, K-Nearest Neighbor, and Naïve Bayes.

Main Results:

• The proposed two-stage hybrid model demonstrated improved success rates in identifying cancer cells.
• Decision Tree, K-Nearest Neighbor, and Naïve Bayes classifiers achieved promising accuracy using the developed framework.
• The method achieved higher accuracy compared to existing approaches in colon cancer classification.

Conclusions:

• The two-stage feature selection approach prior to classification enhances cancer cell identification accuracy.
• The developed hybrid framework offers a promising avenue for improving colon cancer diagnosis.
• This research provides insights that could advance colon cancer treatment and drug discovery.