SpineDAO Chronos Project Spec

Chronos: Representing and Revisiting Overlooked Historical and Traditional Spine Medical Insights with Modern Knowledge Background

Background: In the 1960s, amidst a global malaria crisis, Chinese scientist Tu Youyou systematically organized and analyzed ancient Chinese medical texts, uncovering a 4th-century reference to sweet wormwood (Artemisia annua) as a fever remedy. By rigorously testing this traditional knowledge with modern methods, she discovered artemisinin—a transformative antimalarial drug that earned her the 2015 Nobel Prize in Physiology or Medicine.

Her success hinged on structuring disparate historical insights into a testable framework, a process Chronos aims to automate and scale for spine surgery using AI and decentralized knowledge graphs.

Problem Statement: Spine research holds a wealth of overlooked clinical observations from historical Western medical texts and traditional medicine, yet while some of these insights have made it to modern science, not all have been integrated or explored.

Current knowledge systems fail to bridge past and present effectively, hindering the generation of novel hypotheses. Inspired by Tu Youyou’s systematic approach, Chronos seeks to unlock these hidden insights by finding the ideal format for structuring historical and modern spine surgery knowledge.

Within Spine Archives—a partnership between SpineDAO, the French Académie Nationale de Chirurgie, and Preprints.io—participants will access historical Spine texts from the Académie to build an agentic knowledge representation and mining system, judged by a medico-technical team on its scientific and technical merits.

Challenge: Create Chronos—a system consisting of a decentralized semantic network and agentic hypothesis generation engine that mines overlooked clinical observations from historical spine surgery and traditional medicine texts, integrating them with modern scientific knowledge through an optimized knowledge graph structure.

This structure, the central piece of the project, must integrate established frameworks like the Hypothesis and Evidence Ontology (HypE), The Open Research Knowledge Graph (ORKG), or Unified Medical Language System (UMLS) with OriginTrail’s Decentralized Knowledge Graph (DKG) to systematically represent and link past and present medical knowledge. Develop an autonomous hypothesis generation engine atop this network to propose innovative, testable research directions for spine surgery.

1. Core Element—Optimized Knowledge Graph Structure: Design and implement the ideal knowledge graph structure for medical hypotheses, integrating established frameworks—e.g. Hypothesis and Evidence Ontology (HypE), The Open Research Knowledge Graph (ORKG), or Unified Medical Language System (UMLS)—with OriginTrail’s DKG. This structure must: ○ Formalize historical clinical observations and traditional spine surgery knowledge into machine-readable, refutable hypotheses ○ Link these to modern scientific concepts and evidence ○ Enable traceability from past observations to present research opportunities ○ Facilitate cross-domain connections and hypothesis testing

2. Comprehensive Knowledge Cartography: Populate the knowledge graph with overlooked clinical observations from historical spine surgery texts (provided by SpineDAO) and traditional medicine, alongside modern spine healthcare domains, highlighting forgotten theories, overlooked clinical observations and procedures, and emerging links.

3. Autonomous Hypothesis Generation Engine: Develop an autonomous engine that leverages the optimized knowledge graph to generate testable novel hypotheses for spine surgery. Each hypothesis must include contextual evidence (e.g., historical observations, modern studies, translation nodes).

4. Decentralized Knowledge Integration: Use OriginTrail’s DKG plugin to write findings to a shared "SpineGraph," with BioAgents automatically adding nodes and edges as they process historical and modern literature.

5. Bonus Knowledge Preservation Tasks: ○ OCR Digitization: Implement OCR using libraries like RolmOCR to digitize undigitized Spine Archives texts yourself. ○ Redundant Storage: Agents to store digitized documents on various centralized and decentralized services for preservation and accessibility. Test dataset and prototype example: The challenge will provide digitized and undigitized versions of a French historical Spine medical texts from the Académie Nationale de Chirurgie for participants to use as an example of their knowledge graph structure and hypotheses generation engine. Additionally, Chronos includes a prototype example of the complete workflow—OCR digitization, an optimized knowledge graph using HypE and novel hypotheses generated from a historical text. This example will be provided to participants as an example to improve upon.

Evaluation Metrics:

Submissions will be evaluated by a medico-technical team on the following criteria:

● Knowledge Graph Quality: Scientific rigor, structure, and effectiveness of the chosen framework implementation for representing and linking past and present knowledge ● Scientific Accuracy: Correctness of extracted historical, traditional, and modern spine surgery information and connections ● Research Quality: Rigor, novelty and sophistication of the generated hypotheses ● Cross-Domain Integration: Ability to connect knowledge across historical observations, traditional medicine, and modern domains ● Implementation Quality: Code quality, documentation, and reproducibility ● User Experience: Quality of visualizations and system usability ● Bonus Task Completion: Quality of OCR digitization and effectiveness of redundant storage Desired Outcomes: ● An optimized knowledge graph structure integrating HypE, ORKG, or UMLS or another appropriate medical knowledge representation framework with OriginTrail’s DKG, comprehensively representing historical and modern spine surgery knowledge ● An interactive visualization system for exploring the integrated spine surgery knowledge landscape ● An automated research engine generating novel, testable hypotheses with supporting evidence from the knowledge graph ● Documentation on system architecture, data sources, and knowledge graph ontology ● (Optional) Implementation of bonus tasks: OCR digitization and redundant storage