February 28, 2019

AI for Rare Diseases: Putting Tech to Work for Small Populations

Photo of a child and mom with face paint

Today is the 11th annual Rare Disease Day, a worldwide observance originally instituted by the European Organisation for Rare Diseases to raise awareness for diseases that affect relatively few people.

Ironically, rare diseases add up. One estimate is that more than 350 million people in the world – that’s more people than live in the United States – or about 1 in ten people worldwide, have one of about 7,000 diseases that are considered rare.  More people than those affected by depression, the most common disability. More than cancer and AIDS combined.

The problem with rare diseases, however, is right there in the name. Each condition is so rare that healthcare becomes extremely challenging. It’s often difficult to diagnose, develop treatments, or find support as a patient when you’re dealing with something that affects so few people. Ninety-five percent of rare diseases lack an FDA-approved treatment. But this is where today’s technology and artificial intelligence (AI) can help.

Disease Awareness and Diagnosis
Some rare diseases present clearly and are diagnosed at birth – in fact, half of all people with rare diseases are children. But many have symptoms that are not straightforward and are easily confused, which is why the average time to diagnosis for a person with a rare disease is eight years. By this time, some of its effects can be irreversible.

General practitioners aren’t going to have thousands of uncommon conditions in mind when they see their patients and have probably not even heard of many of them. And in some cases, a patient might go to one healthcare professional (HCP) with one complaint, and another with a different complaint.

Putting all the clues together and flagging the likelihood of a rare disease is something at which AI can excel. Algorithms can combine information from disparate medical records, pulling together highly distributed symptoms, and continuously calculate the probability, and level of confidence, of any number of diagnoses that can be presented to an HCP for consideration in real time.

Clinical Trial Recruitment
Access to clinical trials can be highly sought-after by patients seeking new treatments – and this is the facet of trials that is often covered in the media. However, to test potential medications, researchers need to find enough of the appropriate patients and convince them to join a controlled clinical trial. In the case of rare diseases, this can truly be like finding the proverbial needle in a haystack. It’s a time-consuming and expensive process that involves finding the very few diagnosed patients that fit stringent trial-inclusion protocols.

In many cases, clinical trial recruitment remains archaic: fliers are taped up in hospital elevators in the hope that patients may pass through and notice. AI offers enormous potential to improve this fundamentally vital part of medicine, using data to prequalify patients and target the likeliest audience with contextually relevant messages.

Programmatic Advertising
Similarly, even once patients are diagnosed, and a treatment has been approved, it’s hard to make sure the right people know that a new option exists for them. Programmatic advertising can help. And AI-powered campaigns, when planned and executed by experts who can ensure both regulatory compliance as well as the most effective targeting available, can change the game for rare-disease brands. For example, a recent Intouch effort – a collaboration between teams of data science, media, accounts, project management, and analytics experts – was extremely successful, cutting a rare-disease brand’s cost per site visit from $306 to $39.

Patient Support
If you have a common condition like depression, diabetes, heart disease or even cancer, you almost definitely know someone else with it too. Rare disease patients are another story. They might go their entire lives without meeting another person with their condition. When you consider the life-changing implications a disease can have, particularly a genetic one (as most rare diseases are), you begin to see the enormous value that peer support can have for a patient. From empathizing with the emotional implications of a diagnosis, to giving advice on the practical effects a disease can have on day-to-day life, peer support can be literally life-saving.

Social media is an amazing vehicle that has had well-publicized consequences, both positive and negative. But the ability of AI-powered social algorithms to connect people who would never otherwise meet is unprecedented in human history. Today, even if their disease limits them physically, patients with rare diseases can be involved in empowering communities that allow them to connect and advocate for themselves.

Empowerment is the real power in AI. It’s helping us do more, better, than we’ve ever been able to before. It can provide better diagnoses sooner. It can help researchers find patients for clinical trials. It can help patients find the right treatments for them – and it can help them find each other. Focusing on rare diseases is the right thing to do to help hundreds of millions of underserved patients and AI can help us do it better.