Oxford PharmaGenesis, a leading HealthScience consultancy, and Trace Labs, the software company behind the Decentralised Knowledge Graph (DKG), have announced a partnership to pioneer a collaborative, agentic clinical data ecosystem. The initiative aims to accelerate scientific discovery, improve transparency, and unlock new ways for artificial intelligence to access and use trustworthy medical data.
“The collaboration begins with a pilot, which will link together publicly available information from multiple medicines produced by a global pharmaceutical company. It will create the blueprint for rapid expansion to additional contributors through a structured, incentivized data-sharing program that will form a domain-specific paranet within the OriginTrail DKG.”
Publicly Available Information in the Pilot
The pilot will utilize publicly available data to ensure feasibility, transparency, and alignment with open science principles. Likely data types include:
- Clinical Trial Registrations: Data from platforms like ClinicalTrials.gov, including trial designs and objectives.
- Clinical Study Report Synopses: Publicly shared clinical study report summaries from regulatory submissions.
- Peer-Reviewed Publications: Articles published in scientific journals detailing trial results and insights.
- Regulatory Summaries: Documents from agencies like the FDA or EMA, outlining drug approvals and safety profiles.
- Trial Protocols and Statistical Plans: Publicly disclosed study methodologies and analytical frameworks.
- Real-World Evidence: Aggregated data from real-world studies, such as patient outcomes or drug performance.
How would it work?
There are no specifics on functionality, but here is an outline of how it could potentially work:
- Contribute Data: Pfizer uploads trial data for Ibrance (palbociclib) to the clinical data ecosystem – or “paranet”. The data is formatted in the standardised way (to be determined) and cryptographically signed (ownership is verified).
- Read Data: Researchers and healthcare professions submit information requests through a dedicated DKG portal. AI agents read the data to produce a plain-language summary.
- TRAC Rewards: Node operator receives TRAC tokens for its contribution (maintaining the node). More on this later.
- Expansion: AstraZeneca and Takeda join, contributing data for Tagrisso and Entyvio, respectively, scaling the paranet’s scope and utility.
Likely Candidates for the Global Pharmaceutical Company
The identity of the global pharmaceutical company involved in the pilot is undisclosed, but based on Oxford PharmaGenesis’ client relationships, transparency commitments, and the scope of publicly available data, the following are the top candidates:
- Pfizer
- Rationale: A major client of Oxford PharmaGenesis (e.g. Ibrance), Pfizer is a major contributor to clinical trial data sharing initiates. Its commitment to Open Pharma and extensive portfolio in oncology and vaccines make it a strong fit.
- Relevant Data: Public trial data for Ibrance.
- AstraZeneca
- Rationale: A key client for Oxford PharmaGenesis (e.g., Tezepelumab/Tezspire), AstraZeneca is a leader in open science, particularly in oncology and respiratory drugs. Its U.K. base aligns with Oxford PharmaGenesis’ headquarters.
- Relevant Data: Trial data for Imfinzi, Tagrisso, or Tezspire.
- Takeda
- Rationale: A long-term client (e.g., for Entyvio), Takeda has a global presence and participates in data-sharing initiatives, particularly in gastroenterology and oncology.
- Relevant Data: Trial data for Entyvio or Adcetris.
- Novo Nordisk
- Rationale: A client for Oxford PharmaGenesis (e.g., for Icodec), Novo Nordisk is a leader in diabetes and obesity research with significant public trial data.
- Relevant Data: Trial data for Ozempic or Icodec.
Other Potential Candidates: Novartis, Roche, and Ipsen are also viable due to their global reach and public data-sharing practices.
A “structured, incentivized data-sharing program”
What do they mean by this? While specific details of the program are not fully disclosed in the provided context, we can infer its likely structure, components, and mechanisms based on the goals of the initiative. The “structured, incentivized data-sharing program” would be a carefully designed framework to encourage pharmaceutical companies, regulators, and other stakeholders to contribute their clinical trial knowledge to the Decentralised Knowledge Graph.
Key Components of the program
- Structured Framework for Data Contribution:
- Data Types: The pilot will focus on publicly available clinical trial data, such as trial registrations (e.g., from ClinicalTrials.gov) and clinical study report (CSR) synopses. In later phases, it may include patient-level data under controlled access.
- Standardised Formats: Data will be formatted into machine-readable structures compatible with the OriginTrail DKG, using standards like JSON-LD or RDF to ensure interoperability and discoverability.
- Contribution Process: Contributors (pharmaceutical companies, academic institutions, etc.) will use secure, intuitive tools (likely web-based interfaces or APIs) to upload clinical data to the DKG.
- Pilot as a Blueprint: The pilot, linking public data will establish templates for data ingestion, validation, and structuring. For example, trial data could be formatted into a standardised schema, including trial IDs, endpoints, and results.
- Incentivisation: To encourage participation, the program will offer incentives to pharmaceutical companies and other stakeholders. Likely incentives include:
- Reputation and Transparency: Contributing to a trusted, open knowledge pool enhances a company’s reputation for transparency.
- Access to Insights: Contributors gain access to aggregated, anonymised insights from the DKG, such as trends in clinical outcomes or real-world evidence, which can inform R&D or market strategies.
- Token-Based Rewards: The OriginTrail ecosystem uses the TRAC token to incentivise data sharing in its network. Contributors may receive TRAC tokens for uploading verified data.
- Collaborative Opportunities: Contributors may gain priority access to partnerships with other stakeholders in the paranet, such as academic researchers or patient groups, fostering collaborative research.
- Operational Framework:
- Onboarding Process: Pharmaceutical companies will be invited to join the paranet through a formal onboarding process, likely managed by Oxford PharmaGenesis due to its extensive network. The pilot will onboard one major player (e.g., Pfizer, AstraZeneca, or Takeda) to test the system, with a clear roadmap for scaling to additional contributors.
- Verification and Security: The DKG uses cryptographic technology to ensure data integrity, with each contribution cryptographically signed to verify ownership. Smart contracts may automate validation.
- Access Control: While the pilot focuses on public data, the program will include safeguards for sensitive data. Role-based access controls and encryption will protect proprietary or private information.
- AI Agents: The paranet will enable AI agents to consume and produce knowledge, generating outputs like plain-language summaries, scientific reports, or visual explainers.
- Governance Model: A governance framework, possibly overseen by a consortium including Trace Labs, Oxford PharmaGenesis, and key contributors, will define rules for data quality, contribution criteria, and dispute resolution. This ensures the paranet remains trusted and scalable.
- User-Facing Outputs:
- The program will produce tailored outputs for various stakeholders:
- Researchers: Structured datasets for meta-analyses or systematic reviews, accessible via API or DKG queries.
- Healthcare Professionals: Summaries of clinical evidence for decision-making.
- Patients: Plain-language summaries making trial results understandable and actionable.
- Public: General insights into medical advancements, combating misinformation with verified data.
- The program will produce tailored outputs for various stakeholders:
What is a ‘paranet’?
A paranet is essentially a sub-network. In this case, a sub-network for clinical trial data: interconnected nodes in a web. The DKG is a network itself. A paranet is a network within the larger DKG network. Other paranets exist in the DKG e.g supply chain. Only members of the paranet can get access or contribute. There are likely to be different tiers of membership:
Founders: Top-level, those that have developed the paranet
Contributor Pharmaceutical companies/Individuals: Parties contributing data to the paranet
Readers: Those reading the data (via AI Agents). They could be other pharmaceutical companies, clinical consultancies, regulators, researchers, healthcare professionals.
What to learn more about the DKG?
Click here to learn more about the DKG and sign up to video guides and on-site training.
Resources
