Stephen Kingsmore has dedicated his career to using genomic sequencing to diagnose critically ill infants with mysterious diseases. Among his primary challenges has always been the numbers game—how to provide genome sequencing for the tens of thousands of children annually who could benefit from the diagnostic tool when each test is costly and time-consuming.
Enter artificial intelligence (AI). “The key to rapidly expanding access to genome sequencing is to think about using machines to do many of the tasks that a machine can actually do better than a human,” says Kingsmore, M.D., DSC, president and CEO of Rady Children’s Institute for Genomic Medicine (RCIGM) at Rady Children’s Hospital in San Diego. “This is just the start of that. This is one of the early areas where artificial intelligence could really be a game-changer.”
Automation boosts speed, scale
In a study recently published in Science Translational Medicine, Kingsmore and his colleagues outline the processes by which AI addresses some of the limitations of traditional genome sequencing methods. Using automation to aid genome sequencing can provide numerous benefits:
- Speed. Automation-assisted genome sequencing provides a diagnosis within a day, as opposed to an average of 10 days previously. That rapid turnaround can be critical to the implementation of life-changing treatments for critically ill infants.
- Scale. Kingsmore says as few as 5% of the estimated 80,000 to 100,000 neonatal intensive care unit and pediatric intensive care unit patients annually who need genome sequencing receive it, but the new methods allow for several tests to be run simultaneously on a single sequencing instrument. “That’s huge,” Kingsmore says. “That means we finally have a method that can realistically scale to meet the needs of 100,000 kids per year."
- Access. With cloud-based AI automating the analysis and interpretation of data, it’s now possible for intensive care unit physicians around the world to use genome sequencing in treating their patients.
- Manpower. The highly specialized experts needed to conduct genome sequencing are in very short supply—according to the American Board of Medical Genetics and Genomics, there were fewer than 1,600 available nationwide as of 2017. AI does the heavy lifting of manual data review, freeing up human experts to focus on critical analysis and difficult cases.
“The power of answers”
In addition to the quantitative benefits of genome sequencing, Kingsmore says the testing also provides less tangible—if not equally as important—benefits for patients and their families.
For the roughly 50% of genome sequencing tests that yield no diagnosis, there’s a sense of relief. “They’re so pleased because we’re telling them that with the best test we have available today, your baby has a healthy genome,” Kingsmore says. “Your baby doesn’t have a genetic disease. From an insurance company’s standpoint, the test was a bust. But what we’re hearing from the half of parents who don’t get an answer is that it’s actually great news to know their child has had a genome sequenced and we found nothing.”
Kingsmore adds that the parents who are often most appreciative of the testing are those who receive the worst possible news—their child’s condition is untreatable.
“Parents are appreciative when you actually give them an answer,” Kingsmore says. “They can start to get their head around what this means for their future, their child’s future. They come away with less depression and anxiety and can make difficult decisions about their child. That’s not something that comes out in the study manuscript, but that is part of the secret of why this is going to change health care—it’s the power of answers.”