BRIDGE: A Computational Literature-Based Discovery Engine for Identifying Convergent Mechanisms and Generating Testable Multi-System Disease Hypotheses - Applied to Alzheimer's Disease - Convergent Evidence from Twenty-Four Independent Research Runs with Dual-Model Cross-Validation
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
BRIDGE (Biomedical Research Intelligence for Discovery and Gap Exploration) is a computational literature-based discovery engine designed to identify mechanistic connections across disciplinary boundaries in complex disease research. The system harvests peer-reviewed papers from thirteen global sources, embeds them using SciBERT into a vector database, clusters by semantic similarity, and applies a dual-model synthesis architecture using Gemini 3.1 Pro Preview and Claude Opus 4.6 independently to identify lost correlations - mechanistic links that exist across the literature but have not been previously synthesized into unified models. Applied to Alzheimer's disease across twenty-four independent research runs analyzing 3,293 unique papers, BRIDGE produced four primary trans-disciplinary hypotheses: the Pathogen-Metabolic Nexus connecting metabolic vulnerability to catastrophic pathogen response; the Neuro-Thyroid Ferroptosis Bridge linking shared NCOA4-mediated ferritinophagy mechanisms across neurodegeneration and thyroid autoimmunity; the Tau Lactylation Vise proposing a complete molecular chain from insulin resistance through tau K677 lactylation to ferroptotic cell death; and the Axoglial Death Spiral identifying myelin as the primary substrate for iron-catalyzed lipid peroxidation in neurodegeneration. Dual-model cross-validation produced four CONVERGED findings independently confirmed by both models: the Gut-Tryptophan-TLR4 axis driving glymphatic collapse; PKM2-mediated microglial metabolic paralysis; ferroptosis as primary executioner via Nrf2/GPX4 pathway failure; and APOE4-mediated repression of TFEB-driven lysosomal biogenesis. A dedicated treatment protocol run produced a computationally derived three-phase intervention framework whose central finding - that triggering cellular clearance while macro-drainage remains blocked causes secondary neuroinflammation - offers a potential explanation for the consistent failure of single-target Alzheimer's clinical trials. We propose these hypotheses and the intervention framework as candidates for experimental validation, and present BRIDGE as a replicable open methodology demonstrating that meaningful cross-disciplinary hypothesis generation including dual-model cross-validation is accessible beyond traditional credentialed research environments.
Fields of Study & Subjects
Alzheimer's disease research and treatments
Biomedical Text Mining and Ontologies
Computational Drug Discovery Methods
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 30 March 2026
- Language
- EN
- Document type
- Preprint