Artificial intelligence (AI) is beginning to find its place in rare disease research. With small populations, fragmented registries, and pressure on timelines, it is no surprise that sponsors are exploring new tools to close evidence gaps.
One idea that has emerged is the creation of “synthetic patients” – AI-generated profiles designed to stand in for real patient voices or experiences. It sounds efficient. But in reality, it risks losing the very things that matter most in rare disease: trust, cultural nuance, and lived experience.
Why AI is Attractive in Rare Disease
The barriers are well known:
- Patient numbers are small, making it hard to run large, powered trials.
- Evidence is scattered across registries, case reports, and advocacy groups.
- Families cannot wait years while sponsors debate feasibility.
Recent research shows why AI is being explored here:
- Synthetic datasets have been created for conditions such as cystic fibrosis, sickle cell disease, and Duchenne muscular dystrophy (Orphanet Journal of Rare Diseases, 2024).
- Generative models such as Onto-CGAN are being tested to simulate data for diseases not represented in existing registries (npj Digital Medicine, 2025).
- In silico trials are recognised as a way to model trial scenarios when patient numbers are limited.
These approaches show potential – but they are best seen as complements, not substitutes.
The Risks of “Synthetic Patients”
Creating digital stand-ins for real people brings clear risks:
- Bias: Sparse or incomplete training data can amplify existing gaps.
- Ethics: The Lancet Digital Health (2025) cautions that synthetic data should only ever supplement, not replace, genuine patient evidence.
- Trust: Patient advocates themselves have been clear – “AI doesn’t equal a real person.”
These points matter. Under EU Regulation 2021/2282, the Joint Clinical Assessment requires structured input from patients and clinical experts. Synthetic profiles do not meet that bar.
Where AI Can Add Value
There are areas where AI can help, without undermining credibility:
- Evidence gap mapping – highlighting where outcomes such as fatigue in Duchenne or caregiver burden in MPS are missing.
- Accessibility – drafting plain-language summaries that can then be reviewed and improved with patients.
- Traceability – logging and linking patient insights so reviewers can see how input has shaped evidence.
This is where AI belongs: reducing duplication, helping teams organise data, and ensuring that contributions are not lost in the process.
Cultural and Contextual Nuance
Rare diseases are not experienced in the same way everywhere. Access to specialists, routes to diagnosis, and support for families all vary across Europe. So do languages and ways of expressing symptoms. Some countries have well-organised patient groups, others do not.
Synthetic profiles flatten these differences into an “average patient.” That may look tidy on a screen, but it fails to reflect the diversity that regulators – and families – expect to see.
The RAREready™ AI-Enhanced 4D Framework
At RAREready™, we take a different view. AI is part of the answer – but only when combined with authentic patient engagement. That’s why our 4D Framework is designed to support every stage of JCA readiness, from early planning through to submission.
- Discover: Establish governance and identify risks early, including comparator choices and patient evidence gaps.
- Define: Plan immediate actions, align teams, and set milestones.
- Develop: Co-create and integrate patient narratives and evidence in ways that are compliant and traceable.
- Deliver: Produce submission-ready outputs — narrative sections, lay summaries, traceability logs — aligned with Regulation (EU) 2021/2282 and HTACG 2025 guidance.
It is not about shortcuts. It is about making sure every contribution counts, and that sponsors can move forward with both speed and credibility.
A Balanced Path
AI will play a growing role in rare disease research. The key question is how.
Synthetic patients may offer efficiency, but they risk credibility. The better path is to use AI where it adds value – mapping gaps, supporting traceability, making information more accessible – while keeping patients themselves at the centre.
Takeaway: In rare disease, AI should strengthen, not replace, the patient voice. For sponsors preparing for EU JCA, synthetic shortcuts are not an option. A balanced, patient-centred strategy is.