WASHINGTON D.C. –For years, people with rare diseases faced a long and painful diagnostic odyssey. In many cases, it lasted years. For some families, it lasted a lifetime. Now, that path may shrink to just days, thanks to a major FDA policy change backed by artificial intelligence and advanced biology models.
On February 23, 2026, the U.S. Food and Drug Administration FDA introduced a new “Plausible Mechanism Framework.” Under this policy, certain gene therapies and RNA-based treatments may win approval without the usual large randomized clinical trials.
Instead, the FDA can rely on strong biological proof, supported by AI-based prediction tools, to judge how a therapy works. That shift could open the door for patients with conditions so rare that standard studies were never realistic.
This article looks at why old FDA rules often failed patients with ultra-rare diseases, how the new framework uses AI to fill key evidence gaps, what it could mean in practice, how experts are reacting, and where this policy may lead next.
What Are Rare and Ultra-Rare Diseases?
A rare disease affects fewer than 200,000 people in the United States. Ultra-rare diseases affect even fewer people, sometimes only a small number of patients around the world.
- More than 7,000 rare diseases have been identified, yet only about 5% have approved treatments.
- Many are caused by genetic changes and lead to severe, life-threatening symptoms early in life.
- Patients often visit many doctors before getting the right diagnosis, which can delay care for years.
These disorders create a serious problem for traditional drug development because there often are not enough patients for standard clinical trials.
Micro-question: Why are rare diseases so hard to fit into the FDA approval process?
The usual system depends on large randomized controlled trials to show safety and effectiveness. For ultra-rare diseases, finding enough patients can be unrealistic or impossible. At the same time, costs rise and expected revenue stays low, so many companies stay away. In addition, the course of these diseases can differ from one patient to another, which makes data harder to interpret.
The Limits of the Traditional FDA Process
Most drugs must go through:
- Several trial phases with hundreds or thousands of participants
- Randomized, placebo-controlled study designs
- Long-term safety tracking
That model often breaks down for rare diseases.
- Patient groups are too small, so randomization may be unfair or impossible.
- Development costs are high, while timelines stay long.
- As a result, only a small share of rare diseases ever get treatments.
For patients with ultra-rare genetic disorders, the old system often left no clear route forward.
Micro-question: How long does the diagnostic odyssey usually last?
Many patients wait five to seven years, or longer, for a diagnosis. Some never get one at all. During that time, disease progression may continue without treatment, and families are left with fear, confusion, and few answers.
The Plausible Mechanism Framework
FDA Commissioner Marty Makary, MD, MPH, and other agency leaders announced the new framework as a different path for certain therapies. It allows approval based on one well-controlled study , often with only a small number of patients, plus strong evidence showing how the treatment works at the biological level.
Main requirements for eligibility:
- The disease must have a known genetic, cellular, or molecular cause.
- The therapy must directly address that root problem.
- Researchers must understand the natural history of the untreated disease.
- The treatment must show clear target engagement, such as successful gene editing.
- Patients must show major clinical improvement that does not match the untreated course of the disease.
At first, the framework focuses on genome editing, including CRISPR, and RNA therapies such as antisense oligonucleotides. Still, the FDA may broaden it later.
Micro-question: Which treatments could qualify under the new FDA policy?
The main focus is on individualized gene-editing and RNA-based therapies for serious or life-threatening genetic diseases. In some cases, modular product designs may allow one platform to treat several mutations in the same gene.
How AI Helps Build Predictive Biological Evidence
The framework centers on biological reasoning, but AI gives that process much more power. Advanced computer models, in silico simulations, and AI-based prediction tools can help fill evidence gaps when human trial data is limited.
Here are some of the main ways AI supports this shift:
- Predictive modeling, AI can simulate disease progression and likely treatment effects using natural history data, which reduces the need for large control groups.
- In silico methods, newer approach tools, including AI-supported organ-on-chip systems and computational biology models, can predict safety risks and off-target effects.
- Biomarker discovery, machine learning can study genetic and molecular data to identify markers that may predict patient benefit.
- Trial design, AI can help build master protocols and improve analysis of very small datasets.
FDA guidance issued in 2025 and 2026 on AI in drug development also stresses model credibility. That matters because regulators need confidence that these tools are accurate enough for approval decisions.
Micro-question: Can AI take the place of large clinical trials in rare disease care?
Not fully. Still, AI can strengthen the evidence package by producing predictive biological data. It helps support substantial evidence through modeling, mechanism-based proof, and analysis of very small patient groups. That makes approvals possible in cases where the old model offered no path at all.
Real-World Example, Baby KJ and Personalized CRISPR
In 2025, infant KJ Muldoon became the first patient to receive a fully personalized CRISPR gene-editing treatment for CPS1 deficiency, an ultra-rare urea cycle disorder.
- Teams at Children’s Hospital of Philadelphia and partner groups developed the therapy in only six months.
- The treatment was built for KJ’s exact mutation.
- Early results showed major improvement, and that case helped shape the FDA’s new thinking.
KJ’s case showed that highly personalized therapies could be developed quickly and could work in practice. Now, the new framework gives similar treatments a clearer regulatory route.
Micro-question: What changed for Baby KJ after treatment?
KJ showed strong clinical gains that did not match the normal course of his disease. His case became a powerful example of what custom gene therapy might achieve, and it helped influence FDA policy.
What This Means for Patients and Families
The possible benefits are wide-reaching.
- Faster access, patients may get treatment in days or weeks instead of waiting years.
- New hope, people with ultra-rare mutations may finally have a real treatment path.
- Lower long-term costs, smaller trials and AI-supported development may make these therapies more practical to build.
- Personalized care, treatments can be designed for one patient or a very small group, then expanded through master protocols.
Patient advocacy groups such as NORD have welcomed the added flexibility. They say it responds to serious unmet need while keeping safety in view.
Micro-question: How could this change the diagnostic odyssey?
Genetic sequencing, paired with AI-based analysis, can identify harmful mutations much faster. Then, once a diagnosis is clear, the new framework may allow rapid therapy design and review. That could turn years of waiting into a much shorter path to action.
Risks and Safety Checks
Some critics worry that very small studies may not reveal enough about safety at the start.
The FDA says the framework includes several safeguards:
- Post-marketing requirements and commitments for more monitoring and follow-up studies
- The option to withdraw approval if later evidence does not confirm benefit
- A focus on strong target engagement and major, easy-to-see clinical improvement
- Long-term follow-up for gene-editing therapies
The agency also says benefit-risk decisions must reflect how serious the disease is and whether any other treatment exists.
Micro-question: Does skipping large trials weaken safety standards?
No. The framework still calls for strong mechanistic and supporting evidence. In addition, post-approval monitoring continues after a therapy reaches patients, and the FDA can act quickly if concerns appear.
Reactions From Experts and Industry
The response has been strong across the field.
- FDA leaders:Commissioner Makary called the move a “critical step” in adapting regulation to ultra-rare disease care. CBER Director Vinay Prasad described it as a “revolutionary advance in regulatory science.”
- Researchers:Teams involved in Baby KJ’s treatment said the framework could help more patients receive personalized CRISPR-based therapies.
- Industry:Biotech companies expect a wave of applications for custom treatments.
- Advocates:Patient groups say the policy brings regulation closer to what modern biology now makes possible.
Even so, some experts want careful rollout so scientific standards stay strong.
What Comes Next?
Right now, the framework applies to gene and RNA therapies for rare genetic diseases. Still, the FDA has left open the chance to expand the idea to other highly targeted treatments, including some for more common diseases with a clear molecular cause.
AI is also likely to play a larger role over time. As the FDA keeps refining guidance on model credibility and in silico tools, developers may gain faster ways to support approval decisions.
By 2030, some experts expect dozens of additional gene therapies to reach patients through this pathway, changing what rare disease care looks like.
Micro-question: Could this framework reach beyond ultra-rare diseases?
Yes. FDA officials have said the same core ideas could apply to any condition with a clear molecular target and a tailored treatment approach. If that happens, the impact could spread well beyond rare disease care.
Challenges Still Ahead
Several issues still need work:
- The public comment period for the draft guidance will help shape the final version.
- Custom therapies bring manufacturing challenges that may require added flexibility.
- Researchers need stronger natural history databases and clear standards for AI validation.
- Access must be fair across different patient groups and communities.
Because of that, regulators, drug developers, patient groups, and technology experts will need to keep working together.
Conclusion, A New Chapter for Rare Disease Care
The FDA’s Plausible Mechanism Framework, supported by AI-based predictive tools, marks a major change in how rare disease therapies may reach patients. Treatments once seen as impossible to approve for one person or a tiny patient group may now have a real regulatory path.
For families living with rare diseases, that could mean far more than faster paperwork. It could mean quicker diagnosis, treatment built for the actual cause of disease, and a real chance at hope after years of waiting.
This policy does not lower the bar. Instead, it updates the rules to match modern science, so patients are not left out simply because their condition is uncommon.
