Multimodal feature-optimized approaches for cancer classification using microarray gene expression analysis.

Area of Science:

• Genomics
• Computational Biology
• Artificial Intelligence in Medicine

Background:

• Cancer's high recurrence and mortality rates necessitate early and accurate detection.
• Gene expression data is crucial for initial cancer classification but presents challenges due to high dimensionality and small sample sizes.
• Artificial intelligence (AI) methods are increasingly vital for improving cancer diagnosis, survival prediction, and recurrence assessment.

Purpose of the Study:

• To develop precise and effective cancer genomics analysis techniques using advanced computational and analytical methods.
• To present an Artificial Intelligence-Based Multimodal Approach for Cancer Genomics Diagnosis Using Optimized Significant Feature Selection Technique (AIMACGD-SFST) model.
• To enhance early cancer detection and improve patient survival rates through AI-driven genomics.

Main Methods:

• Data preprocessing included min-max normalization, handling missing values, and label encoding.
• The Coati Optimization Algorithm (COA) was employed for optimized significant feature selection.
• Ensemble models, including Deep Belief Network (DBN), Temporal Convolutional Network (TCN), and Variational Stacked Autoencoder (VSAE), were utilized for cancer classification.

Main Results:

• The AIMACGD-SFST model demonstrated superior performance across three diverse datasets.
• Achieved classification accuracies of 97.06%, 99.07%, and 98.55%.
• Outperformed existing models in cancer classification accuracy.

Conclusions:

• The AIMACGD-SFST model offers a precise and effective approach for cancer genomics diagnosis.
• AI-driven feature selection and ensemble deep learning significantly enhance cancer classification accuracy.
• The developed approach holds promise for improving early cancer detection and patient outcomes.