EcoXAI: Autonomous Agentic Ecosystem for Explainable Artificial Intelligence and Biomedical Discovery

Area of Science:

• Biomedical Informatics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Biomedical datasets and knowledge graphs are growing in complexity, creating bottlenecks for researchers seeking actionable insights.
• Existing AI solutions can suffer from hallucinations and workflow fragmentation, hindering reliable data analysis.
• There is a need for autonomous analytical tools that integrate data and biomedical knowledge without compromising accuracy.

Purpose of the Study:

• To introduce EcoXAI, a modular, containerized multi-agent system designed to streamline complex biomedical data analysis.
• To provide researchers with tools that deliver meaningful, actionable insights grounded in both data and established biomedical knowledge.
• To lower the computational barrier for clinical and translational researchers using advanced AI.

Main Methods:

• Developed EcoXAI as an autonomous, execution-driven framework with specialized bioinformatics agents.
• Integrated explicit knowledge graph connections and modern agentic management for transparency and verifiability.
• Structured analysis into explicit pipeline execution stages to avoid workflow fragmentation.

Main Results:

• EcoXAI successfully replaced monolithic AI text interfaces with a specialized agent framework.
• In a drug repurposing study for Alzheimer's Disease, 79 novel drug candidates were identified out of 103 evaluated.
• A hypothesis for the CCR5 antagonist Maraviroc was generated and subsequently supported by existing literature.

Conclusions:

• Knowledge graph-grounded AI agents can accelerate hypothesis-driven biomedical research.
• EcoXAI offers a transparent and verifiable approach to AI-driven biomedical data analysis.
• The EcoXAI system demonstrates significant potential for advancing drug discovery and other biomedical research areas.