Machine learning-based DNA microarray analysis for disease detection using the MICRO-AI framework.

Area of Science:

• Genomics and Bioinformatics
• Artificial Intelligence in Medicine
• Machine Learning for Disease Diagnosis

Background:

• DNA microarrays generate high-dimensional gene expression data, posing challenges for disease pattern identification due to noise and class imbalance.
• Accurate and efficient analysis of microarray data is crucial for automated disease diagnosis and understanding complex biological systems.

Purpose of the Study:

• To develop and validate MICRO-AI, a comprehensive machine learning framework for DNA microarray analysis and automated disease diagnosis.
• To address challenges of high dimensionality, noise, and class imbalance in microarray data.
• To improve classification accuracy and efficiency in identifying diagnostic patterns.

Main Methods:

• Integrated advanced preprocessing techniques including quantile normalization, ComBat batch correction, and KNN imputation.
• Employed attention-weighted adaptive feature selection via recursive feature elimination with cross-validation.
• Utilized a heterogeneous ensemble classifier combining gradient boosting machines, random forests, and support vector machines with adaptive weight optimization.
• Introduced a novel attention-based feature fusion mechanism for dynamic prioritization of discriminative gene expression signatures.

Main Results:

• MICRO-AI reduced data dimensionality by over 99% (from ~20,000 to ~127 genes) without loss of biological significance.
• Achieved average classification accuracy of 96.8% across six benchmark datasets spanning breast, gastric, ovarian cancers, and leukemia.
• Demonstrated superior performance compared to 10 state-of-the-art methods, with 1.2-7.5% higher accuracy and significantly faster training times (average 52.3s).

Conclusions:

• MICRO-AI provides a robust and efficient framework for DNA microarray analysis and automated disease diagnosis.
• The framework's modular architecture facilitates integration with medical informatics systems for clinical deployment.
• MICRO-AI offers a scalable solution for extracting meaningful diagnostic patterns from complex genomic data.