AI drug discovery success inspires a machine-learning startup at the Mayo Clinic
A little over a year ago, Andrew Badley, the chief medical officer and new tech leader at the Mayo Clinic, thought it would be a good time to start testing the waters on artificial intelligence-driven technology for drug discovery. Working with Murali Aravamudan at nference, they put some of the latest tech on data-driven machine learning to use for one of the investigators at the Mayo Clinic who had been doing some hard thinking about a molecular target.
“Literally, in the next couple of weeks we had insights we had not had in the last two years,” recalls Badley. And after they chalked that up as a success, they began to think about the potential of opening AI tech up to all the investigators at the Mayo Clinic.
Now, Badley and Aravamudan are creating a new company, Qrativ, with the sole role of making this AI platform available to Mayo clinicians looking for a machine-assisted approach to dot connecting.
Think of this in terms of the way two scientists can make a breakthrough, they say. Scientist 1, with two ideas comes together with another scientist with a shared notion and a third original idea that’s related to the work. Together, they hatch a new, original theory.
With AI, says Badley, a clinician can take a drug or drug candidate and start asking some questions. What does the drug do? How does the mechanism apply in various diseases? Are there other, better uses for a therapy? If you get a hit, are you looking at the 4th or 5th new drug for a disease that’s already well controlled, or a new entry among the unmet medical needs still on an infinitely long arm? What, ultimately, has the best chance of succeeding?
“The notion of machine learning has been around in drug discovery for awhile,” says Aravamudan, who’ll be running the virtual show at Qrativ for Mayo, with $8.3 million in venture capital from the Mayo Clinic, Matrix Capital Management and Matrix Partners.
In the last few years the field has begun to get crowded. We’re seeing a growing list of new ventures like Numerate, BenevolentAI, Recursion Pharma, or Insilico Medicine in Baltimore step up making the technology available to drug discoverers.
In Qrativ’s case they can include the data sets at Mayo and put it in an AI program that can start drawing lines between drugs, targets and diseases. Take the best ideas into the clinic for proof-of-concept data and then either license out the work to biopharma or start up a new company with it. As you can imagine, there will be various ways of splitting up the IP.
“The day when a machine can automatically ID a drug is still far off,” says Badley. Somebody has to do the hard lab work that lays the foundation for a new drug program. But AI can potentially advance the process by years, eliminating a lot of waste. That’s the hope.
That journey starts now at the Mayo Clinic.