New, cloud-based technologies helped stanch Ebola in Africa and give doctors test results in minutes. Machine learning may help make personalized cures affordable.
When Ebola returned to the Democratic Republic of Congo, doctors weren’t sure what they were facing. Decentralized DNA sequencers let them identify the deadly infectious agent in just 24 hours.
“The devices plug into a laptop computer and cost no more than a mobile phone,” said Dr. Gordon Sanghera, Chief Executive Officer of U.K.-based Oxford Nanopore Technologies. The devices connect to a genomic database stored in the cloud. In contrast, the mainframe-based DNA sequencers that dominate the industry are too heavy (half a ton) and too expensive ($100,000 to $1 million each) to be used in remote locations.
Next-generation, decentralized sequencers could also be used to identify and treat hospital-acquired pneumonia, which is rapidly mutating and becoming antibiotic-resistant, Sanghera said.
He delivered these comments in early November at the Healthcare Innovation and Disruption panel at JPMorgan’s Advisor Exchange in Paris, where entrepreneurs discussed new technologies disrupting healthcare today. JPMorgan Chase Chief Executive Officer Jamie Dimon arrived early and stayed throughout the session to learn about ways to improve healthcare while cutting its cost, the goal of the firm’s joint venture with Amazon and Berkshire Hathaway.
“Our healthcare system is fundamentally broken because it’s way too expensive,” Cary Gunn, Chief Executive Officer of California-based Genalyte, declared at the panel. One way to save money: Run blood tests in a machine in a doctor’s office that provides results while the doctor is still with the patient, typically within 15 minutes. Having data in hand makes doctors more efficient, Gunn said. It also improves outcomes. Patients diagnosed with diabetes, for example, are more likely to change behavior if the doctor tells them face-to-face to cut out sugar rather than if the doctor gives them laboratory results and advice on the phone a week later.
Theranos, another California startup, also promised to provide rapid blood test results; it then collapsed amid fraud accusations, making it difficult for Genalyte to raise capital. But Gunn said his technology works. It uses silicon chips that run on light, not electricity, to “see” how light interacts with molecules in a blood sample, and sends the data to a cloud-based library to be analyzed. The results go straight into the patient’s online medical records.
Each chip can test for 128 molecules; it costs only pennies to add an additional chip to the machines, he said. “Our vision is every time a patient touches the healthcare system, we test them for everything.”
Affordable personalized cure?
Personalized cancer cures are famous for their exorbitant costs: In 2017, uniQure withdrew Glybera from the market only two years after launching “the world’s most expensive medicine” at $1 million per dose. But Dr. Ugur Sahin, Co-Founder and Chief Executive Officer of Germany’s BioNTech, said personalized cures are needed to cure cancer—and the cost will come down eventually.
Take melanoma, the deadliest type of skin cancer. Ten-year survival rates may soon rise to 50% from below 15% today, thanks to better diagnosis and treatments. But that means half of all melanoma patients will still die within 10 years of treatment. “Cancer is a very personal disease, based on mutations that are different for every patient,” Sahin said. While “pre-manufactured drugs” may hit 90% or even 99% of the tumor cells, the remaining cells grow and spread, he explained, adding “then the physician has to say, ‘We don’t have any more drugs to give you.’”
Personalized cures are expensive because the manufacturing process is complex, Sahin said. “More than 150 people worked in a series of steps to make this vaccine,” he noted, holding up a vial with an immunotherapy BioNTech created to treat the mutations in a single patient’s tumor.
But machine learning that predicts the most important targets will help drive down costs, he adds. Still, he admitted, “It will take years to reduce the cost, through scale, full automation and digitalization.”