In our new whitepaper AI in Cancer Care: A Responsibility and an Opportunity (July 2026), Silvia Rohr and Poka Cui draw on a stakeholder roundtable held at ESMO AI 2025 to set out how the life sciences industry can help make AI in oncology safe, transparent and genuinely useful for patients and clinicians. Their argument rests on three observations:
- AI in cancer care is already here. It is shaping how patients and clinicians experience oncology today, from patient information to imaging.
- Delay is not neutral. The absence of trusted AI pushes patients toward unvalidated tools and leaves new therapies stranded in fragmented care pathways.
- Collaboration makes trusted AI possible. The life sciences industry cannot deliver trusted AI alone; working with clinical and technology partners, it can help validated AI reach patients faster and more safely.
Why cancer care needs a connective layer
Cancer care has outgrown traditional models. Clinicians must make decisions using large amounts of data, across multiple steps in the pathway, and often under time pressure. At the same time, staff shortages, administrative burden and fragmented systems with scattered data and disconnected workflows slow care down.2
AI has arrived within this reality. It will not solve everything, but it already acts as a connective layer that helps organise the complexity: linking data, supporting decisions and connecting the steps of the pathway that today operate in isolation.
Patients consult AI before they consult a doctor
The first opinion many patients seek now comes from an algorithm.3 Patients arrive at consultations with AI-generated information and look to their clinician for further guidance. This is shifting expectations and roles on both sides of the consultation, and the healthcare system has to adapt to this new reality. AI is also embedded in clinical practice itself, most visibly in medical imaging, where it already works alongside radiologists.1
For the life sciences industry, the implication is direct: support clinicians with safe, clinically validated tools, because patients now expect their care team to engage knowledgeably with AI-informed questions.
Innovation is not enough: a therapy creates value only when patients receive it
The value of a treatment increasingly depends on the pathway around it. Even when effective treatments exist, many patients still face delays or inconsistent follow-up, and delay in cancer treatment measurably affects outcomes.4 Success will depend on how well the steps of care (diagnosis, planning, treatment and monitoring) are connected, not just on new drugs or devices.
This is the opportunity for AI: acting as the connector that carries scientific innovation through to the patient. The industry’s role is to work with clinicians and technology partners on evidence-based AI that guides patients through the pathway correctly, so every therapy delivers its full real-world value.
If validated AI tools are missing, unvalidated tools get used
AI adoption hinges on trust and clear guardrails, yet there are inherent trade-offs between generality, accuracy and simplicity.5 Fast, cheap and broad tools may scale quickly, but often at the expense of reliability. Without rigorous validation and transparency around these trade-offs, unproven AI will fill the gap.
Clinicians must be able to understand how a system works, what data it uses, and how much confidence to place in its outputs.6,7 Otherwise, hesitation and unsafe shortcuts will replace informed adoption. The industry’s role is to work in partnership to make AI tools explainable and to demonstrate how they function in order to earn trust.
Four foundations for trusted AI
The whitepaper identifies four foundations for building trust in clinical AI:6,7
Transparency
Show how the AI works at every step. In practice this means training data disclosure, feature importance (which factors mattered most), saliency maps (where a model focused in an image), attention-based weighting (which data influenced decisions) and example-based reasoning (similar cases and how changes would alter outcomes).
Integration
Be compatible with existing workflows and systems. AI that sits outside the clinical workflow adds burden instead of removing it.
Stepwise adoption
Begin with low-risk applications and scale only when tools are proven safe and effective.
Security
Comply with applicable healthcare data and AI regulations from the outset.
The industry’s role in the AI ecosystem
The life sciences industry has the knowledge and tools to advance AI adoption and improve cancer care, but it cannot deliver trusted AI alone. The whitepaper describes five areas of industry action. Collaboration: work with stakeholders across the ecosystem for safe AI adoption.
Validation: prioritise clinical validation studies so AI tools meet safety and efficacy standards before scaling.
Education: accompany AI adoption with clinician education programmes.
Integration: invest in data standards and interoperability frameworks for seamless data exchange.
Regulatory guidance: provide regulators with best-practice examples to inform regulatory frameworks.
Around the industry sits a wider ecosystem, each with a distinct role: technology companies developing AI capabilities that become clinically useful; research institutions generating independent evidence; healthcare institutions integrating AI into care and generating real-world insights; patient organisations ensuring AI is designed around patient needs and trust; and governments and regulators establishing policies that enable safe adoption and innovation. The life sciences industry’s position is to connect that ecosystem, turning AI capability into clinical impact and trusted care.
About this whitepaper
AI in Cancer Care: A Responsibility and an Opportunity (Vintura, July 2026) was written by Silvia Rohr, Principal at Vintura, and Poka Cui, Partner at Vintura. It builds on insights gathered during a stakeholder roundtable at ESMO AI 2025 with participants from across the oncology and AI ecosystem, including Belén Algaba (Daiichi Sankyo), Yin Cai (AWS), Christiane Höper (Synagen) and Christian Ludwigs (Aigora).
Download the full whitepaper (PDF)
Would you like to discuss what trusted AI adoption means for your organisation? Get in touch with our team. Or sign up to receive updates on new publications and insights.
Frequently asked questions
What does Vintura’s whitepaper on AI in cancer care cover?
The whitepaper examines how AI is already shaping cancer care, why delayed adoption carries clinical and commercial risk, and what the life sciences industry must do, together with clinical and technology partners, to make AI validated, transparent and trusted for patients and clinicians.
How are patients already using AI in cancer care?
Many patients consult AI tools before seeing a doctor, arriving at consultations with AI-generated information and looking for further guidance from a knowledgeable clinician. AI is also embedded in clinical practice itself, most visibly in medical imaging.
Why is delaying AI adoption in oncology risky?
Delay is not neutral. Without trusted, validated AI, patients turn to unvalidated tools, and new therapies remain stranded in care pathways too fragmented to deliver their full value. Validated, transparent AI closes both gaps.
What does it take to build trust in clinical AI tools?
Four foundations: transparency (showing how the AI works at every step), integration with existing workflows and systems, stepwise adoption that starts low-risk and scales when proven safe and effective, and security through compliance with healthcare data and AI regulations.
What role should the life sciences industry play in AI adoption?
The industry should connect the ecosystem: collaborating with clinical and technology partners, prioritising clinical validation, supporting clinician education, investing in data standards and interoperability, and providing regulators with best-practice examples, turning AI capability into clinical impact and trusted care.
References
- Rohan Krishna et al. (2025). Artificial Intelligence in Radiology: Augmentation, Not Replacement. Cureus. doi: 10.7759/cureus.86247
- EFPIA/Vintura (2023). Innovation for sustainable cancer care: Addressing urgent workforce shortages.
- Ipsos (2024). Can we rely on generative AI for healthcare information?
- Hanna, T. P. et al. (2020). Mortality due to cancer treatment delay: systematic review and meta-analysis. BMJ. doi: 10.1136/bmj.m4087
- Health Union (2025). Connected Health Experiences and Perceptions of AI Survey.
- American Medical Association (2024). Augmented Intelligence Development, Deployment, and Use in Health Care (PDF).
- World Health Organization (2024). Ethics and governance of artificial intelligence for health: guidance on large multi-modal models.
