Pasar al contenido principal

Call for proposals: AI tools to transform evidence synthesis

Cochrane is looking for AI tool developers to propose tools for consideration for the pilot initiative

Image
Artificial intelligence


Cochrane invites proposals from AI tool developers to participate in a pilot initiative aimed at transforming how we produce and share high-quality evidence syntheses. This is an exciting opportunity to collaborate with a globally trusted organization in health research and contribute to the future of systematic reviews.

The background

Cochrane is a global leader in evidence synthesis, producing trusted systematic reviews that inform health decisions worldwide. As the volume and complexity of health research grows, we are exploring how artificial intelligence can help us scale our work while maintaining the highest standards of rigour, transparency, and integrity.

Cochrane recently published a joint position statement on the responsible use of AI in evidence synthesis with Campbell Collaboration, JBI, and the Collaboration for Environmental Evidence. This collaborative effort is an important step in shaping how AI is integrated into the production of high-quality, trustworthy research.

We are now launching a pilot programme to evaluate AI tools that can support or enhance key stages of the evidence synthesis lifecycle; from literature search, screening and data extraction to updating reviews and disseminating findings.

Cochrane’s objectives for this pilot include:

  1. Test tools that could be recommended for review authors at the end of the process;
  2. Collaborate with AI tool developers to help them develop tools that are aligned with RAISE recommendations and fit-for-purpose;
  3. Establish standards for evaluation methods, accuracy thresholds, and validation approaches when introducing AI systems to systematic review processes;
  4. Build capacity to support future evaluations by developing datasets, infrastructure, and methodologies. 

What we’re looking for

We welcome proposals from developers of AI tools that address one or more of the following areas across different types of, and approaches to, systematic reviews:

  • Study identification and screening
  • Data extraction (including structuring and consistency checks)
  • Risk of bias assessments
  • Evidence interpretation and synthesis
  • Translation and accessibility
  • Review updating and automation
  • Knowledge dissemination and visualization
  • Please note, for this pilot we are likely to prioritize AI tools focused screening and data extraction.

What the tools should demonstrate

  • Alignment with Cochrane’s mission, vision and values:  
  • Adherence to the Responsible AI use in evidence SynthEsis (RAISE) initiative
  • Technical feasibility: interoperability with existing platforms (e.g., RevMan), scalability, and usability
  • Explainability and transparency: ability for humans to clearly understand and interpret decisions made by the AI tool, to facilitate replicability and accountability.
  • Privacy and security: compliance with data protection and copyright standards 

Pilot scope

Cochrane central staff will complete an internal assessment. AI tools that best meet the criteria (see below) will be assessed by members of the joint AI Methods Group.  

Cochrane will select two-to-three tools for the initial round of piloting. These will be tested in a defined pilot setting, which may include:

  • Application to a group of Cochrane reviews to automate updates
  • Use in a sandbox environment with synthetic or real data
  • Evaluation within Cochrane processes using a diagnostic test accuracy framework
  • Collaboration with Cochrane experts and authors for support, feedback and refinement 

Selection criteria

Please provide details about the following:

  1. Tool usability and alignment with Cochrane expectations:
    • The tool aligns with Cochrane’s mission and ethos of transparency, reproducibility, and research integrity, as well as facilitates or ensures adherence to Cochrane’s methodological standards.
    • The tool can be effectively used by review authors as part of the systematic review process (e.g., easy to use interfaces or similar), and demonstrates practical value in improving efficiency, relevance and / or accuracy.
    • Clear audit trails of decisions within the AI tool to support explainability of the system.
       
  2. Alignment with RAISE  
    • Commitment from the tool developers to support the aims of RAISE and willingness to collaborate and adapt their tool as needed, including evidence of commitment to ethical, transparent, and fit-for-purpose AI development (e.g., compliance with data protection and copyright standards).
    • Appropriately populated RAISE framework for assessing AI tools or systems (see below), which includes key information about the tool, its strengths, limitations and potential biases, and evidence of performance and validation.
       
  3. Evaluation standards and validation approaches
    • The tool supports or enables clear evaluation methods (e.g., reproducible metrics, benchmarking), and the AI tool development team provides mechanisms or documentation for performance thresholds and validation approaches.
    • Willingness to collaborate and support Cochrane staff, authors and others during the pilot (see timeline below).
       
  4. Capacity building and infrastructure
    • The tool development team provide details of the technical overview and system architecture, as well as (data) interoperability with different platforms (e.g., RevMan and other evidence synthesis platforms), any impact for publications on the Cochrane Library, and scalability potential.
    • The tool creator is willing to contribute to dataset development, infrastructure, or methodologies for future evaluations, to facilitate long-term capacity building within the evidence synthesis community.

Please also detail any costs associated with using the AI tool as part of a pilot.

Any proposals that do not include enough information for Cochrane to be able to evaluate against all these areas will not be shortlisted.

Timeline

  • Submission deadline: 31 December 2025
  • Review and selection: end of Q1 2026
  • Pilot launch: beginning of Q2 2026
  • Pilot duration: 3 months

Get involved

If you’re building AI tools that could help shape the future of evidence synthesis, we want to hear from you.

Submit your proposal hereRead more on the RAISE initiative

Nuestro uso de las cookies

Utilizamos las cookies necesarias para que nuestro sitio funcione. También nos gustaría establecer cookies analíticas opcionales para ayudarnos a mejorarla. No estableceremos cookies opcionales a menos que las habilite. El uso de esta herramienta establecerá una cookie en su dispositivo para recordar sus preferencias. Puede cambiar sus preferencias sobre las cookies en cualquier momento haciendo clic en el enlace "Configuración de las cookies" en el pie de cada página.
Si desea información más detallada acerca de las cookies que utilizamos, consulte nuestra página sobre Cookies.

Aceptar todo
Configurar