The human immune system is incredibly complex and incredibly smart. Your immune system knows when you’re getting sick, sometimes days, months or even years before the problem is diagnosed.
If we were able to read the tiny, microscopic signs of our immune systems, we could catch and treat diseases like ovarian or pancreatic cancer before they become deadly. An early screening tool like that could save millions of lives a year — and Microsoft is working to make it a reality.
The tech giant announced Thursday that it is partnering with Seattle biotech company Adaptive Biotechnologies on a years-long project to build a universal blood test that can screen for dozens or even hundreds of diseases at a time, all by decoding the information in your immune system.
As part of the deal, Microsoft is investing an undisclosed sum in Adaptive and is offering the company cloud computing and machine learning services. The combined value of those two is in the “hundreds of millions of dollars,” Adaptive CEO Chad Robins told GeekWire. The project is part of Microsoft’s Healthcare NExT program, which aims to put Microsoft technologies to work on solving problems in healthcare.
If the project is a success, it would be a radical change to how we diagnose diseases from cancer to multiple sclerosis and would mark a new era for health technology.
The idea is simple in theory: Use the body’s natural red flag system to tell when someone is sick.
“Fundamentally, nature or evolution is more advanced in some ways than we’re ever going to be,” Adaptive Co-founder and CTO Harlan Robins, also Chad Robins’ brother, told GeekWire. “The system our bodies have for detecting and getting rid of disease is called the adaptive immune system, and it’s going to be better at detecting disease than we’ll ever be.”
All the information our immune systems know is encoded into the genes of immune cells like T-cells and B-cells that float around in our blood. Adaptive uses technology called next-generation sequencing to read the genes of those cells and create an immune profile made up of billions of data points.
Somewhere in that profile, buried in mountains of data, are the signs of any disease that’s brewing.
“It contains all the information about every disease you have in your system or have ever had in your system, but we haven’t been able to interpret that yet,” Harlan Robins said. “The goal of this project is to really learn how to interpret that information in real time.”
That’s where another new technology steps in: Machine learning. Microsoft will put its artificial intelligence clout behind the data Adaptive has gathered, building a blood test that learns how to connect the dots between the genetic markers on immune cells and diseases.
Unlike current diagnostics, which look for one specific disease, the test could screen for broad swaths of diseases at once. Chad Robins, the Adaptive CEO, compared it to an X-ray machine.
“An X-ray machine can be used to diagnose a broken wrist but it can also be used to diagnose a broken leg,” he said. In the same way, the test could be used to diagnose diseases as different as pancreatic cancer and multiple sclerosis, an autoimmune disease.
In the perfect scenario, this test would become routine. At every annual checkup, every person would be screened for diseases through a lightning-fast analysis of the markers in their immune system. It could even help us predict how people will respond to diseases and treatments.
“Imagine a world in which an ‘X-Ray of the immune system’ actually exists,” Peter Lee, Microsoft CVP of AI and research, wrote in a blog post about the project. “This would open new doors to predictive medicine, as a person’s immunological history is believed to shape their response to new pathogens and treatments in ways that are currently impossible to explore. The impact on human health of such a universal blood test that reads a person’s exposure and response to disease would be, in a word, transformational.”
Adaptive will focus their initial work on three classes of disease: Diseases that are often diagnosed in very late stages, like pancreatic and ovarian cancer; Autoimmune diseases that are typically hard to diagnose, like multiple sclerosis; and infectious diseases that can remain in a person’s system and reoccur, like chronic Lyme disease.
Harlan Robins said the biggest challenge of the project may well be the regulatory hurdles. Currently, diagnostics are only approved to diagnose one condition at a time, meaning the test must be re-approved for each application.
The insurance system also poses a hurdle — after all, the test won’t be successful if no one will pay for it.
But if it is successful, this test could change the way we think about health and introduce a proactive mindset to preventing and treating diseases. It’s also yet another example of the way that artificial intelligence is changing the healthcare and biotech industries, often taking advantage of data so voluminous it was previously almost unusable.