Artificial intelligence has enormous potential to transform cancer care. But moving from promising technology to meaningful change in hospitals and care pathways requires much more than high-performing algorithms.
It requires clinicians who understand the realities of cancer care. Patients and citizens who can challenge assumptions and bring lived experience into the conversation. Policymakers and regulators who understand the conditions for responsible deployment. Health system experts who know what implementation looks like in practice. And technology specialists who can translate emerging possibilities into safe, interoperable and usable solutions.
This is why PATH has been working from its earliest stages to understand who needs to be involved in shaping the responsible implementation of AI across the cancer pathway.
Mapping the ecosystem around AI and cancer care
During the first six months of the project, PATH carried out a structured stakeholder mapping process coordinated by SHINE. Rather than simply compiling a list of organisations, the work examined the stakeholder ecosystem through three complementary perspectives: the sector in which stakeholders operate, their position across the cancer care continuum, and their relationship with AI-enabled healthcare.
The first dimension uses a Penta-Helix approach, bringing together Public Administration, Health and Care, Business and Industry, Knowledge and Academia, and Civil Society and Citizens. The aim is to avoid viewing healthcare innovation from a single institutional or disciplinary perspective and instead recognise the interconnected roles of policy, research, healthcare, innovation and society.
The second applies a life-course perspective to the cancer pathway. Stakeholders were considered from awareness and prevention, screening and diagnosis through treatment, survivorship, recurrence and palliative care. This is particularly important because AI should not be seen as a collection of isolated applications: its potential value lies in how it can support people and professionals across a connected care journey.
Finally, stakeholders were mapped according to their relationship with AI, including development and innovation, clinical implementation, governance and regulation, data infrastructure, end-user representation, and training and capacity building. Together, these dimensions provide PATH with a systematic way of identifying the expertise required for responsible AI implementation in oncology.
What did we find?
An internal mapping survey revealed considerable scientific and technical strength. Approximately 45% of respondents represented Knowledge and Academia, while the consortium also showed strong expertise in AI development and innovation, data infrastructure and digital systems, clinical implementation, training, governance, and ethics and social sciences.
Across the partner organisations, partners identified approximately 70 available experts, demonstrating substantial multidisciplinary capacity within the consortium. However, the exercise also revealed an important imbalance: expertise was particularly concentrated in technical and academic domains. The largest identified gaps were in governance and regulation, clinical implementation and patient representation. Partners therefore prioritised external involvement from oncology healthcare professionals, regulatory and policy experts, patient representatives, and implementation and health-system specialists. The message was clear: PATH does not primarily need more technical expertise around the table. It needs complementary expertise that can help translate technological innovation into solutions that are clinically meaningful, responsible, acceptable and implementable.
From stakeholder mapping to an Expert Group
These findings have directly informed the establishment of the PATH Expert Group, bringing together approximately 30 experts across six complementary profiles: clinical oncology and healthcare professionals; governance, policy and regulatory experts; health systems and implementation experts; patient, citizen and societal stakeholders; industry and sustainability stakeholders; and AI and interoperability support experts.
Importantly, the composition intentionally places greater emphasis on clinical, patient, regulatory, health-system and implementation perspectives, exactly the areas where the stakeholder analysis showed that complementary knowledge could add the greatest value to the consortium’s existing expertise.
And this is not an Expert Group designed simply to observe the project from the sidelines.
Members will participate throughout PATH through co-creation workshops, interviews, questionnaires, focus groups, consultations and structured validation activities. They will help identify implementation barriers and facilitators, refine oncology use cases and patient journeys, contribute to user-centred development, interpret validation findings and support recommendations for the responsible implementation of AI in oncology. A multi-round Delphi study will also provide a structured mechanism for building consensus across the different stakeholder perspectives. The first plenary Expert Group meeting is planned for September 2026, marking the beginning of this next phase of engagement.
Because responsible AI is a collective endeavour
The early PATH stakeholder work reinforces a simple but important principle: AI in cancer care cannot be implemented responsibly through technology alone. Its success will depend on whether innovation fits clinical practice, responds to patient priorities, works within health systems, meets regulatory and ethical expectations, and can be sustainably integrated into real care pathways.
By identifying not only the expertise PATH already has, but also the voices it needs to strengthen, the project is building a multidisciplinary community that can help bridge the gap between AI innovation and responsible implementation. Because the future of AI-enabled cancer care is to be clinically useful, ethically grounded, implementable, and shaped with the people who will ultimately make it work.