Welcome back to Healthy Innovations! 👋

This Saturday, February 28, is Rare Disease Day – a global moment of solidarity observed across 106 countries for the 300 million people living with a rare disease, and the families supporting them.

It felt like the perfect week to dig into just how fast this space is moving. From a landmark AI study published in Nature this month to gene therapies reaching patients in record time, there is genuinely a lot of hope to talk about.

Let’s take a look!

In a study published in Nature this month, an AI system called DeepRare did something that would have seemed implausible a few years ago: it outperformed experienced specialist physicians at diagnosing rare diseases.

The researchers tested DeepRare against five rare disease physicians, each with over a decade of clinical experience, using 163 real patient cases from Xinhua Hospital in Shanghai. Both the AI and the physicians received identical clinical information. The physicians had access to search engines and reference materials.

DeepRare “won” – identifying the correct diagnosis as its top prediction 64.4% of the time, compared to the physicians' 54.6%.

That gap matters enormously.

Rare diseases affect more than 300 million people worldwide. The average patient waits more than five years for a correct diagnosis, and in one large survey, 25% of patients reported eight or more consultations before receiving a confirmed answer. Every year of delay means missed treatment, irreversible progression, and a family living in diagnostic limbo.

DeepRare isn't a single model. It's a multi-agent system built on large language models (LLMs), integrating more than 40 specialized tools and connecting to medical literature, rare disease databases, and real-world case repositories in real time. When a patient's information is entered, whether as free-text clinical notes, standardized phenotype terms, or raw genomic sequencing data, the system generates a ranked list of diagnostic candidates, each accompanied by a transparent reasoning chain linking every inference step to verifiable medical sources.

The system uses a three-tier architecture: a central LLM-powered host with memory coordinates the process, specialized agent servers handle phenotype and genotype analysis and knowledge retrieval, and a broad layer of external data sources provides up-to-date medical evidence. A self-reflection loop re-checks the system's own hypotheses before producing a final output, reducing errors and hallucinations.

DeepRare has already been deployed as a web application diagnostic copilot for rare disease physicians.

Ten rare disease physicians reviewed 180 of DeepRare's reasoning chains across eight datasets and agreed with 95.4% of the evidence citations, confirming that the AI wasn't just producing plausible-sounding output, but genuinely grounded reasoning.

Across more than 6,400 cases spanning nine datasets and 14 medical specialties, drawn from clinical centers in Germany, the United States, and China, DeepRare consistently outperformed every comparison group, covering 2,919 rare diseases in total. When genomic sequencing data was added alongside clinical phenotypes, diagnostic accuracy jumped further, with correct top-1 identification reaching 69.1% in the Xinhua Hospital dataset versus 55.9% for the leading bioinformatics tool.

ARPA-H, the U.S. Advanced Research Projects Agency for Health, launched its RAPID program, Rare Disease AI/ML for Precision Integrated Diagnostics, with an explicit mission to collapse the diagnostic timeline. The program is building a large curated dataset of longitudinal rare disease patient data, designed to train and validate AI diagnostic tools at population scale, with direct-to-patient tools that can work outside specialist centers.

"By leveraging AI, we can expand access to rare disease expertise and greatly reduce time to diagnosis, from years to months or even days," said RAPID Program Manager Scott Gorman. "AI-enabled support tools allow us to sift through the 'haystack' of patient data more efficiently and pinpoint the 'needles' of rare diseases."

Faster, more accurate diagnosis matters most when there's something to do with the answer.

The story of Elly Krueger shows what becomes possible when AI-enabled diagnosis, coordinated advocacy, and accelerated gene therapy development converge at the speed patients actually need.

In February 2024, Elly was diagnosed at eight months old with NEDAMSS, Neurodevelopmental Disorder with Regression, Abnormal Movements, Loss of Speech, and Seizures, a progressive neurological condition caused by mutations in the IRF2BPL gene. Only a very small number of cases have been identified worldwide. There were no approved treatments and no established path forward.

Her parents launched Elly's Team immediately and connected with a multidisciplinary team of researchers, clinicians, and regulatory advisors to pursue a gene therapy approach. Safety studies, drug manufacturing, and FDA regulatory review ran simultaneously rather than sequentially. On April 3, 2025, Elly became the first child to receive an IRF2BPL gene replacement therapy, just 14 months after her diagnosis. One month post-treatment, the therapy was safe and well tolerated.

"In the future," said Michelle Krueger, "another family will sit in the hospital and receive the same diagnosis, but their doctor will tell them there is a path to treatment."

Elly's story sits inside a much larger momentum shift. Several dozen cell and gene therapies have now received FDA approval, with renewed activity across multiple rare disease indications.

In April 2025, Abeona Therapeutics won FDA approval for Zevaskyn, a cell-based gene therapy for recessive dystrophic epidermolysis bullosa. The Phase 3 VIITAL trial showed that 81% of wounds treated with a single surgical application achieved 50% or more healing at six months, compared to 16% of control wounds.

Also in 2025, the FDA approved the first gene therapy for Wiskott-Aldrich syndrome, with the approval going to Fondazione Telethon ETS, a non-profit organization, showing that the path from laboratory to patient doesn't require a large commercial sponsor.

Pricing remains a structural challenge. Some approved gene therapies have reached several million dollars per patient, rational economics for diseases affecting only tens of patients annually, but a framework that puts treatments out of reach for most health systems. Platform manufacturing approaches are being explored as a route to lower costs, but distribution of progress remains deeply uneven.

Rare diseases have always punched above their weight in medical innovation.

Early clinical successes in genomic sequencing, mRNA technologies, and gene therapy all emerged from rare, single-gene disorders before spreading to cancer and cardiovascular disease. The tools being validated today for rare disease diagnosis will eventually reach the broader healthcare system.

DeepRare's researchers made this explicit in their Nature paper: the system's consistent performance across 14 medical specialties suggests its potential as a decision support tool not just for rare disease specialists, but for non-specialist physicians who encounter these conditions without the context to recognize them.

Rare Disease Day is a reminder that the 300 million people living with rare conditions are not a niche concern. They are the edge of where medicine is going, where the hardest diagnostic problems get solved first, and where the tools that solve them eventually benefit everyone.

⚡ Novocure treats cancer with electric fields rather than drugs or radiation. Founded in 2000 by Dr. Yoram Palti, the Switzerland-based company delivers low-intensity alternating electric fields through wearable, skin-mounted arrays that disrupt cancer cell division.

The therapy, called Tumor Treating Fields (TTFields), is delivered via portable devices like Optune, supplied through a lease and consumables model.

Already approved for glioblastoma, malignant pleural mesothelioma, and non-small cell lung cancer, Novocure secured a significant new FDA approval in early 2026 for locally advanced pancreatic cancer via Optune Pax. With more than 4,400 active patients worldwide and a robust R&D pipeline spanning brain, lung, and abdominal cancers, Novocure is advancing from a single-indication business into a multi-indication oncology platform.

🏥 The Pitt's AI reality check. HBO's medical drama The Pitt is doing something TV rarely does well: making a genuinely nuanced argument about AI in hospitals. Season two follows Dr. Trinity Santos, a resident drowning in patient charts, who gets nudged toward AI transcription software by a tech-enthusiastic attending.

The software mostly works – until a surgeon comes storming in over errors that could have led to wrong patient care. Rather than going full "AI bad" on us, the show lands on a sharper point: AI transcription can speed up charting, but it can't fix an understaffed ER or an overcrowded waiting room. Sometimes "more productive" just means burning out faster while double-checking a robot's homework.

Thank you for reading the Healthy Innovations newsletter!

Keep an eye out for next week’s issue, where I will highlight the healthcare innovations you need to know about.

Have a great week!

Alison ✨

P.S. Healthcare is evolving at an unprecedented pace, and your unique insights could be invaluable to others in the field. If you're considering starting your own newsletter to share your expertise and build a community around your healthcare niche, check out beehiiv (affiliate link). There's never been a better time to start sharing your knowledge with the world!