Too rare for care: How AI is helping solve the rare disease paradox

With a smartphone, mGene’s AI technology can analyze facial features to identify subtle indications of genetic syndromes

It’s one of medicine’s tragic paradoxes that, collectively, rare diseases have significant impact on human health. The World Health Organization defines conditions affecting fewer than 1 in 2,000 people as rare, and it’s estimated that over 300 million people worldwide currently live with one of the 7,000 conditions classified in this way. That’s six times as many as have been diagnosed with cancer in the last 5 years.

Adding to the issue, patients diagnosed with rare conditions often find themselves on the wrong side of the cost-benefit analysis that’s applied to funding decisions. Resources tend to flow to research and care for conditions with a larger patient population. Low public awareness and a lack of medical studies exacerbate the challenges through the difficulty of diagnosing conditions, which can lead to sufferers being excluded from care. A key contribution that the cloud and AI can make to healthcare outcomes involves rebalancing this numbers game. AI is helping to transform genomics and DNA sequencing, so researchers can scale their work globally, share insights through the cloud, and develop innovative approaches for deeper understanding of patients’ experiences.

The inherent challenges of diagnosing and treating rare disease

Some of the better-known examples of rare diseases include motor neuron disease, cystic fibrosis, Duchenne muscular dystrophy, and hemophilia. As with many rare diseases, these four conditions have one crucial thing in common. They’re caused by genetic variations that can manifest – in a wide range of different ways, at different stages. In all, an estimated 80% of rare diseases have a genetic component, and this helps to explain both why they’re relatively rare and why they can present diagnostic challenges. Because these conditions are related to genetic variations rather than external pathogens and environmental factors, they often fall outside of traditional paths to diagnosis.

Not all rare diseases are inherited, and those resulting from rare bacterial or viral infections, autoimmune responses, or sporadic genetic mutation come with similar challenges. Affecting individuals rather than populations makes these diseases less visible, and they generate much less of a public sense of urgency because they don’t occur in epidemics or pandemics. And so many of those affected by rare diseases are children, who may have difficulty communicating their symptoms, which can contribute to diagnostic complexity and can cause great distress to patients and their networks of support.

Taking a granular view of rare disease through genomics

Greater understanding and ongoing research are helping to illuminate the value of genetic testing in helping to diagnose many rare conditions. Genomics England works with Amazon Web Services (AWS) and AWS Partner Illumina to embed genomic analysis into the diagnostics journey and help transform the speed with which suspected cases of rare conditions can be confirmed. Genomics England’s 100,000 Genomes Project laid the foundations for whole genome sequencing for patients with suspected rare conditions through the NHS Genomic Medicine Service (GMS). The NHS GMS, which is supported by Genomics England, has sequenced more than 100,000 genomes and made the NHS the first national health system in the world to offer whole genome sequencing as part of routine care.

The difference that accurate diagnosis can make

For those affected by rare conditions and their families, the impact of ready access to diagnosis through genomics may help provide important insights. For one mother, Mel, the NHS GMS led to the discovery that two of her children, including one previously diagnosed with autism and dyspraxia, had an ultrarare neurodegenerative condition, caused by a variant in the DHDDS gene. Although there were only 59 documented cases of this condition worldwide, Mel was able to reach out to specialists working on it for suggestions of specific vitamins that could help support management of tremor symptoms.

In a feature on the Genomics England website, she explained at the time that the children’s “tremors have reduced by 20–30% since they started taking them.” However, she’s also clear that support managing symptoms is only the start. As founder of the charity Cure DHDDS, she’s focused on raising funds for research, holding scientific conferences, and building an international patients registry. “We juggle so many balls because we need to act quickly,” she says. “The aim is to find a drug that can slow down the disease whilst we wait for tomorrow’s treatments, such as ASO or RNA therapies.”

Enabling treatments to target gene expression

Antisense oligonucleotides (ASO) and ribonucleic acids (RNA) therapies are treatments that use specific molecules to help regulate disease-causing genes by changing the way the genes are expressed. Cloud computing may help lower barriers to developing these therapies for rare diseases by democratizing access to computing power and generative AI capabilities and providing highly secure, federated datasets to help enable development of novel treatments and support drug discovery efforts. These technologies may also help transform the economics of rare disease therapies, potentially making treatments viable even for conditions affecting smaller populations.

AI and the cloud may help bring other scaling benefits to initiatives tackling rare diseases, as well. AWS has worked with the National Library of Medicine’s National Center for Biotechnology Information in the United States to make the Sequence Read Archive (SRA), one of the world’s largest repositories of genome sequencing data, freely accessible from Amazon Simple Storage Service (Amazon S3) using the AWS Open Data Sponsorship Program. The SRA includes approximately 40 petabytes of data that include about 40 million run accessions on the Open Data Sponsorship Program. By gaining access to the program, researchers worldwide can locate and retrieve sequencing data from those experiments, helping to enable smoother global collaboration and support research efforts.

Detecting the early signals of rare conditions with AI-assisted analysis

One of the greatest challenges involved in diagnosing and treating rare diseases is the difficulty decoding the experiences of the very young children who make up a significant share of sufferers. The Children’s National Hospital in the United States, one of the first recipients of a new $10 million AWS funding program to support research into pediatric health and rare diseases, has built AI technology that can analyze smartphone camera images to assist clinicians in identifying subtle changes in newborns’ facial features. Through this technology, they can detect rare genetic disorders. In many cases, these signals may be identified later in a child’s development, when preventive care is less effective. However, when identified earlier with AI, they can save children from years of misdiagnosis. Instead, they receive the treatment and support they need from the start.

Rady Children’s Hospital and its Institute for Genomic Medicine is one of the first recipients of the AWS Imagine Grant Children’s Health Innovation Award, which supports organizations accelerating innovation in children’s health through advanced cloud services. The hospital is using large language models (LLMs) to help further accelerate diagnosis and help make genomics testing available to the children that need it as soon as possible.

“Leveraging AI and cloud computing is a critical step as we work towards making genetic testing more equitable, inexpensive, and widely available,” said Matthew Bainbridge, PhD, Supervising Research Scientist at Rady Genomics, a nonprofit research institute at Rady Children’s Hospital San Diego. “Working with AWS allows us to utilize their expertise in AI and cloud computing to improve the speed and access to pediatric genetic testing, ultimately shortening a child’s diagnostic odyssey.”

Giving rare disease patients the recognition and support they deserve

For millions of people with rare diseases, the experience often involves a lengthy journey that can cause significant challenges for them and their loved ones. Until now, healthcare systems have faced constraints in funding, data, and understanding needed to properly identify these conditions and address them with the appropriate urgency their life-changing impact deserves. Although this absence of support might feel like indifference to patients, the real obstacle has been technological limitations. By continuing to fill this gap, AI and cloud computing may help improve the experience of living with a rare disease, which can help to give these conditions the attention they require.

Medical Disclaimer

This blog discusses AWS cloud technology solutions and their potential applications in healthcare settings. Any references to healthcare use cases, clinical workflows, or medical applications are provided as examples of how technology may be utilized and should not be interpreted as medical guidance or clinical recommendations.