Structured insights: AI-powered biomedical research leverages massive knowledge graph

Researchers have created a groundbreaking knowledge graph called iKraph that transforms biomedical literature into structured data capable of powering automated discoveries in healthcare. This innovative approach successfully predicted repurposed drugs for COVID-19 treatment early in the pandemic, with a third of its recommendations later validated through clinical trials. The achievement represents a significant advancement in using AI to extract actionable insights from the overwhelming volume of scientific publications, potentially accelerating drug discovery and treatment development for various conditions.

The big picture: A team led by Yuan Zhang has built iKraph, a comprehensive biomedical knowledge graph that won first place in the 2022 LitCoin Natural Language Processing Challenge by successfully converting unstructured scientific literature into usable structured data.

• The system extracts information from all PubMed abstracts at human-expert level accuracy while significantly exceeding the content available in manually curated public databases.
• The researchers enhanced the knowledge graph’s completeness by integrating relation data from 40 public databases and high-throughput genomics data.

Why this matters: Converting the rapidly growing volume of scientific literature into actionable intelligence represents one of the most significant challenges in biomedical research.

• Manually curated databases can’t keep pace with publication growth, creating an information bottleneck that slows scientific progress.
• iKraph’s ability to transform unstructured text into structured knowledge enables rapid insight generation that would be impossible through traditional literature review.

Key achievement: The system demonstrated real-world impact through COVID-19 drug repurposing efforts from March 2020 to May 2023.

• Using an interpretable, probabilistic-based inference method, iKraph identified approximately 1,200 candidate drugs in the first four months of the pandemic.
• One-third of the drugs discovered in the first two months were later supported by clinical trials or PubMed publications, validating the system’s predictive capabilities.
• These results would be extremely difficult to achieve without comprehensive understanding of existing literature, highlighting iKraph’s unique value.

What’s available: The researchers have made the technology accessible to the scientific community through multiple platforms.

• A cloud-based platform (https://biokde.insilicom.com) provides academic users access to the structured data and associated analysis tools.
• The system’s code is available on GitHub (https://github.com/myinsilicom/iKraph), with data accessible through a Zenodo repository (https://doi.org/10.5281/ZENODO.14846820).
• An additional platform at https://www.biokde.com offers expanded functionality.

The big question: iKraph’s success raises important implications for the future of scientific discovery.

• Can automated systems consistently identify valuable connections in literature that human researchers might miss?
• Will this technology accelerate drug discovery timelines beyond the emergency conditions of a global pandemic?
• How might this approach transform other scientific fields struggling with information overload?