Apply AI: Piloting AI-based image screening in medical centres

According to guidance “Participation in EU calls with ownership and control restrictions” V2.0 (01.01.2026) (guidance-on-participation-in-eu-calls-with-ownership-and-control-restrictions_en.pdfPDF), “If ineligible non-EU country/entity/national control is already known at proposal submission, you are strongly advised to already prepare the guarantee as part of your proposal (even if the guarantee itself, will be requested from you only later on, during grant preparation)”. Therefore, it is possible to submit the proposal without the guarantee.

Participation in a project funded under the Digital Europe programme does not create an exemption from the Medical Devices Regulation or applicable national regulatory requirements.

The public guidance available on Funding and Tenders Portal - guidance-on-participation-in-eu-calls-with-ownership-and-control-restrictions_en.pdfPDF - states that regarding ownership and control criteria, for natural persons, the analysis will be based on proof of nationality. Furthermore, it states that natural persons with dual nationality (EU and non-EU) will be considered as EU nationals. The same principle will be applied in case of other combinations of dual nationality, one being eligible country and other non-eligible country in the context of specific call conditions. Meaning that eligible country nationality will be taken into account.

An entity established in an eligible country under this call (EU Member States, EEA countries and Switzerland), but with an ownership control assessment as non-EU controlled may still participate in the project, subject to the provision of a security guaranteeWord and its approval by the competent authority of the relevant Member State and by the European Commission.

This call for proposals does not require participation in specific prior EU-funded projects. However, according to section 2 of the call text, specifically the scope subsection, the text states: “Where applicable, they should support innovative initiatives related to cancer, cardiovascular diseases and other noncommunicable diseases and build on the results of successful projects funded under Horizon 2020, Horizon Europe, the Digital Europe Programme, EU4Health, or other relevant Union initiatives supporting innovation, such as the Cancer Image Europe platform”. Therefore, proposals should indicate projects and initiatives that meet the call requirements and are relevant for their proposals. Moreover, relevant policy context, including the Apply AI strategy and the European AI-powered Advanced Screening Centres, is provided in Section 1 (Background) and should inform the preparation of the proposal.

The description of the network and list of participating organisations is published on European AI-powered advanced screening centres webpage.

The secretariat of the network is provided by the European Commission, and its contact details will be shared later with the funded project(s) only.

Security restrictions under Article 12(6) of the Digital Europe Regulation apply to this call for proposals (please see sections 6 and 10 and Annex 2 of the call document for more information). Therefore, “Only the following countries are eligible: EU Member States, EEA countries and Switzerland (participation limited to specific eligible countries). Entities must not be directly or indirectly controlled from a country that is not an eligible country unless the granting authority agrees to allow for exceptional participation on the basis of a guarantee (ownership control restriction)” (section 6).

As stated on page 36 of the call document, “participation in any capacity (as beneficiary, affiliated entity, associated partner, subcontractor or recipient of financial support to third parties) is also limited to entities established in and controlled from eligible countries” and “project activities (included subcontracted work) must take place in eligible countries”.

Please consult the Annotated Grant Agreement model,PDF in particular article 7, for details on the role of the coordinator.

The consortium composition must meet the requirements set out in the subsection ‘Consortium composition’ on page 15 of the call document:

"− minimum 7 independent applicants (beneficiaries; not affiliated entities) from 5 different eligible countries

− at least one industrial partner

The consortium can include public and private entities such as (but not limited to): healthcare providing organisations, medtech industry, SMEs, AI/GenAI and IT solution providers, research organisations, governmental authorities (at national, regional, local level), non-governmental organisation representing key stakeholders, such as patients, healthcare professionals, etc.” Also relevant to consortium composition, in the subsection ‘KPIs to measure outcomes and deliverables’ on page 10, it is stated that proposals must define KPIs including the following: "Number of medical centres in which each cloud-based AI/GenAI system was integrated (at least 5 medical centres from different eligible countries)”. All entities and proposed work must contribute to the objectives, scope, outcomes and deliverables, and KPIs described in section 2 of the call document.

As stated on page 9 of the call document, “The AI/GenAI algorithms and cloud infrastructure/services necessary for the projects must be either provided by the consortia or by a provider from an eligible country.” Please note that due to this call being subject to security restrictions, “Only the following countries are eligible: EU Member States, EEA countries and Switzerland” (please see page 13 of the call document for more information).

In the context of this call, “industrial partner” refers to a private-sector entity with an active operational role in the pilot deployment (e.g. medtech industry, SMEs, AI/GenAI and IT solution providers).

“Further training of the algorithms” (see page 8 of the call document) is within the scope of this call. In addition, as stated on page 8 of the call document, the piloted cloud-based AI systems should be “seamlessly integrated in regular clinical workflows of the care settings involved” and “Interoperability and the need to communicate with the health IT infrastructure (e.g. the electronic health record (EHR) of the patient, Picture archiving and communication systems (PACS), Radiological Information Systems (RIS), etc.) need to be addressed.” Proposals are expected to include costs for personnel to perform these tasks. Please see the subsection “Budget categories and cost eligibility rules” on page 22 of the call document and Article 6.2.A of the Annotated Grant Agreement for more information about personnel costs.

Average personnel costs (unit cost according to usual cost accounting practices) and actual personnel costs are permitted under this call. Please see Article 6.2.A.1 of the Digital Europe programme general model grant agreementPDF for more information about these types of costs.

The Digital Europe Programme Grant Agreement will cover up to 50% of the cost incurred, provided that the general and specific eligibility conditions of Article 6 of the General Model Grant AgreementPDF are met. The remaining co-financing the action is the responsibility of the consortium and may include own resources, financial contribution from third parties or income. In-kind contributions provided by third parties “for free are allowed, but cost-neutral, i.e. they cannot be declared as cost” (section 10, under 'Budget categories and cost eligibility rules').

According to guidance “Participation in EU calls with ownership and control restrictions” V2.0 (01.01.2026) (guidance-on-participation-in-eu-calls-with-ownership-and-control-restrictions_en.pdfPDF), “Guarantees are programme-, call- AND project-specific. You can NOT rely on guarantees given/accepted in the context of another EU programme, call or project.”

The piloting of modified versions of Medical Device Regulation-approved (MDR-approved) AI algorithms is within scope, as is the piloting of modified versions of MDR-approved AI solutions (e.g. a not yet MDR-approved AI algorithm is added to the solution). However, in either case the modifications are required to contribute to the early detection and/or diagnosis of cancer and/or cardiovascular diseases.

The submission of these modified algorithms or solutions would contribute to the KPI on AI algorithms submitted for regulatory approval, and the type of submission (e.g. substantial change) does not need to be specified in the proposal.

Yes, as stated on page 8 of the call documentPDF: “The proposers are expected to choose the types of health data, the AI/GenAI algorithms that are trustworthy and secure, and the medical use cases, with the focus disease areas being cancer and cardiovascular.”

The focus of the call is on algorithms analysing imaging data, as stated in the ‘Objectives’ subsection on page 7 of the call document. However, as stated in the same section, “Additional clinical data of a patient may be used, if available, to support the analysis and prioritisation of the case.”

As stated in the ‘Provisions concerning the project implementation’ subsection on pages 25-26 of the call document, there are three “special IPR obligations linked to restrictions due to security”. More information about these obligations is provided in annex 5 of the Annotated Grant AgreementPDF particularly in the ‘Exploitation of results’ and ‘Transfers and licensing of results’ subsections on pages 358-359.

As stated on page 7 of the call document, “Each project awarded in this action should implement a scalable cloud-based AI/GenAI system running European trustworthy and safe AI/GenAI algorithms that were developed, trained and validated using very large sets of patient data.” In this context, “very large sets of patient data” means sets of data of at least several hundred different patients.

Proposals are required to address at least one of these two disease areas.

There is no standard average timeline, as it depends on several factors, including the time the entity needs to prepare the guarantee, the time required for the national competent authority to review and approve it, where applicable, and the time needed for the European Commission to review and approve it, where applicable. However, the guarantee approval must be completed during the Grant Agreement Preparation Phase and before the Grant Agreement signature. Please note, as stated on page 11 of the call document, the indicative deadline for the Grant Agreement signature is May-June 2027.

The text “at least 5 medical centres from different eligible countries” on page 10 of the call document should be interpreted as “at least 5 medical centres from several different eligible countries”.

Please note that page 9 of the call document states that “The projects must involve medical centres from multiple eligible countries” and that one of the eligibility criteria (on page 15 of the call document) is that “Proposals must be submitted by: minimum 7 independent applicants (beneficiaries; not affiliated entities) from 5 different eligible countries”.

In addition, page 8 of the call document states that “A key element of the piloted system solutions will be their scalability and replicability across various eligible countries and their health systems, enabling consistent and equitable digital service delivery. Any solutions should be compatible with national health systems across eligible countries to allow for easier replication across different contexts.”

AI solutions to analyse imaging data and with clinical impacts including the following are within the scope of this call:

• Improving the early detection and/or diagnosis of cancer and/or cardiovascular diseases.
• Enhancing prioritisation of critical cases.

Please see the objectives section on page 7 of the call document for the full list of clinical impacts.

Based on the objectives and scope of this call for proposals, medical centres can be understood as healthcare providing organisations “performing analysis of medical imaging data they are acquiring while providing healthcare services in their sites using medical imaging devices such as MRI, CT, X-rays, PETs, ultrasound, etc.” (see the objectives subsection on call document page 7). The healthcare providing organisations can be public or private entities (see consortium composition subsection on call document page 15). They are not required to have hospital status.

The piloting of already Medical Device Regulation-approved (MDR-approved) AI solutions is within the scope of the call. However, please note that, as stated on page 10 of the call document, proposals must define key performance indicators (KPIs) including the following two:

• “Number of European AI/GenAI algorithms that were validated by clinicians to generate the necessary evidence for regulatory approval.”
• “Number of European AI/GenAI algorithms submitted for regulatory approval following the validation in real-life scenarios and clinical settings within the projects.”

Due to these two KPIs, each project must pilot at least one AI algorithm that is not yet MDR-approved.

As stated on page 10 of the call document, proposals must define key performance indicators (KPIs) including the following one: “Number of European AI/GenAI algorithms submitted for regulatory approval following the validation in real-life scenarios and clinical settings within the projects.” Due to this KPI, working directly on the Medical Device Regulation certification of European AI/GenAI algorithms is within the scope of this call.

There is no requirement to standardise the approach to validating a cloud-based AI/GenAI system across all medical centres. As stated on page 8 of the call document, “The projects should perform technical and clinical validation of the systems, generating evidence on AI systems’ performance and usability in different clinical settings.” The goal is to generate the type of evidence suitable for regulatory approval and health technology assessment purposes.

There is no requirement that the approach to deploying a cloud-based AI system be harmonised across all medical centres. However, as stated on page 8 of the call document, “Any solutions should be compatible with national health systems across eligible countries to allow for easier replication across different contexts.”

While the call document does not explicitly require that the cloud-based AI system is the same across all medical centres, the pilots are expected to be able to demonstrate that the system is scalable and replicable, as stated on page 8 of the call document, “A key element of the piloted system solutions will be their scalability and replicability across various eligible countries and their health systems, enabling consistent and equitable digital service delivery. Any solutions should be compatible with national health systems across eligible countries to allow for easier replication across different contexts.”

In the context of this call, “screening” is not limited to population screening only. It refers to AI-based image screening which should be understood as “performing analysis of medical imaging data” with the aim to “automatically detect findings of potential clinical significance for a given patient, including incidental findings”. As stated on page 7 of the call document, the AI-based image screening will have clinical impacts including:

• “Improve the early detection and diagnosis of cancer and cardiovascular diseases.”
• “Enhance prioritisation of critical cases, enabling faster decision-making.”
• “Optimize resources in radiology departments, alleviating workload pressures.”

As outlined in the call document, this call for proposals aims to “Improve the early detection and diagnosis of cancer and cardiovascular diseases” (objectives, section 2). The call text also states that “proposers are expected to choose the types of health data, the AI/GenAI algorithms that are trustworthy and secure, and the medical use cases, with the focus disease areas being cancer and cardiovascular” (scope, section 2).

The focus of this call is cloud-based AI systems contributing to the early detection or diagnosis of cancer or cardiovascular diseases by analysing medical imaging data. Use cases in other medical specialities would be in scope if they contributed to the early detection or diagnosis of cancer or cardiovascular diseases.

Costs for application programming interface (API) usage or subscriptions and the rental of cloud computing infrastructure to deploy, evaluate and perform further training of AI algorithms are classified under Category C.3 (“Other goods, works and services”) and are thus eligible as actual costs provided that they contribute to the objectives, scope, outcomes and deliverables, and KPIs described in section 2 of the call document. However, please note that, as stated on page 9 of the call document, “The AI/GenAI algorithms and cloud infrastructure/services necessary for the projects must be either provided by the consortia or by a provider from an eligible country.”

This will be assessed on a case-by-case scenario depending on the impact to the overall proposal.

Please see section 7 ‘Financial and operational capacity and exclusion’ of the call document where the financial capacity description and assessment is explained. Regarding the impact on the prefinancing, this will be assessed during the Grant Agreement preparation phase. There are different scenarios as detailed in the call document, notably:

“If we consider that your financial capacity is not satisfactory, we may require:

− further information

− an enhanced financial responsibility regime, i.e. joint and several responsibility for all beneficiaries or joint and several liability of affiliated entities (see below, section 10)

− prefinancing paid in instalments

− (one or more) prefinancing guarantees (see below, section 10)

or

− propose no prefinancing

− request that you are replaced or, if needed, reject the entire proposal.

For more information, see Rules for Legal Entity Validation, LEAR Appointment and Financial Capacity AssessmentPDF.”