IntelliGenes: Interactive and user-friendly multimodal AI/ML application for biomarker discovery and predictive...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Artificial intelligence (AI) and machine learning (ML) have seen significant advancements across various fields.
• Progress in applying AI/ML to multi-omics data analysis lags due to data complexity and required expertise.
• High-dimensional multi-omics data presents challenges in handling, analysis, and interpretation for biomarker discovery and disease prediction.

Purpose of the Study:

• To introduce IntelliGenes, an interactive, customizable, cross-platform AI/ML application designed for multi-omics data exploration.
• To facilitate the discovery of novel biomarkers and the prediction of various disease types (rare, common, complex).
• To provide a user-friendly interface for both technical and non-technical users to leverage AI/ML in multi-omics research.

Main Methods:

• Developed IntelliGenes, integrating conventional statistical techniques with advanced ML algorithms for enhanced accuracy.
• Implemented a nexus of algorithms that outperform single-algorithm approaches.
• Designed an interactive graphical user interface with three modules: Data Manager, AI/ML Analysis, and Visualization.

Main Results:

• IntelliGenes successfully classifies patients and predicts diseases with high accuracy in tested studies.
• The application integrates data management, customizable AI/ML pipeline creation, and interpretable result visualization.
• Demonstrated superior performance compared to single-algorithm methods through a combination of statistical and ML techniques.

Conclusions:

• IntelliGenes offers a simplified approach for applying AI/ML to multi-omics data, making complex analyses accessible.
• The application empowers users, regardless of computational background, to discover novel biomarkers and predict diseases effectively.
• Its emphasis on interpretable visualizations and ease of use distinguishes it for multi-omics data exploration.