IBM Watson, long touted for its potential in healthcare, is being put to work finding a drug to treat Parkinson’s disease.
The Ontario Brain Institute (OBI) and University Health Networks (UHN) in Toronto, Ontario, is adopting IBM Watson for Drug Discovery to analyze existing data with natural language processing and machine learning tools.
Why Parkinson’s Is So Hard to Treat
Parkinson’s is a progressive disorder of the nervous system. It’s linked to the loss of nerve cells that produce dopamine, a chemical messenger the brain uses to control movement, and it causes symptoms such as tremor, stiffness and slowness. Existing medicines can ease those symptoms, but there’s no treatment yet that stops or reverses the disease.
That’s why researchers are interested in looking at compounds that already exist. If a drug already approved or tested for another condition turns out to act on the biology of Parkinson’s, it can sometimes move into trials faster than a brand new molecule, because a lot is already known about its safety.
Get the TechPortfolio briefing
The month’s fintech and funding news, plus new tools for founders, in one short email.
How Watson Helps
One problem that IBM Watson solves for researchers is processing existing Parkinson’s research, and finding previously overlooked connections within research structures. Watson Drug Discovery’s resources include 31 million sources of data.
No research team can read that much. Natural language processing lets software scan published papers, patents and databases, pick out the genes, proteins and compounds they mention, and map how they’re connected. The output is a list of candidate links for scientists to test in the lab, not an answer in itself.
Dr. Lorraine Kaila, a neurologist at UHN, explains: “The platform gives us the ability to look at connections that researchers might not have found without dedicating weeks or months of time. This includes identifying compounds that we have not previously considered investigating for the treatment of Parkinson’s disease.”
“Drug researchers are challenged by the sheer volume and pace of emerging data,” says Lauren O’Donnell, Vice President at IBM Watson Health Life Sciences. “Watson for Drug Discovery empowers researchers with cognitive tools that will help to speed drug discovery.”
#IBMWatson joins forces with @OntarioBrain & @UHN_News to advance #Parkinsons research. https://t.co/xRtNgBvLM4 pic.twitter.com/VZXxWYmF11
— IBM Canada (@IBMCanada) October 12, 2016
The Cost of Finding New Drugs
According to drug manufacturers’ trade group PhRMA, research for new treatments costs an average of $2.6 billion and takes 10 years. Only 12% of proposed medications make it past a Phase I clinical trial, which tests efficacy and safety.
On average 25 people are diagnosed with Parkinson’s each day, according to the OBI. In Ontario alone, an estimated 285,000 people suffer from some form of neurodegenerative disorder.
Tom Mikkelsen, president and scientific director of the OBI, said: ”This partnership signals the beginning of a new era for neuroscience where researchers can work with data at an unprecedented level of sophistication and speed. We are excited by the impact this could have on people living with Parkinson’s disease.”
What Happened Next
Watson’s broader push into medicine didn’t live up to the early excitement. In 2015 IBM had formed a dedicated Watson Health division, and by mid-2016 it had spent about $4 billion buying four health data companies, according to a 2019 IEEE Spectrum investigation titled “How IBM Watson Overpromised and Underdelivered on AI Health Care.” The same report described how hard real medical records proved for the system, and noted that MD Anderson Cancer Center cancelled one Watson-based project after spending $62 million on it.
In January 2022, IBM agreed to sell the healthcare data and analytics assets of Watson Health to investment firm Francisco Partners. Those assets became a standalone company, Merative.
AI in drug research kept moving forward without Watson at the center. The 2024 Nobel Prize in Chemistry went partly to Demis Hassabis and John Jumper of Google DeepMind for an AI model that predicts the structure of proteins, solving a problem scientists had worked on for 50 years. Knowing a protein’s shape is a key step in designing drugs that bind to it.
What It Means for Health Researchers
The idea behind the OBI and UHN project still makes sense: there’s far more published research than any team can read, and software can help point scientists toward connections worth testing. The lesson from the past decade is that those tools are aids to judgment, not replacements for it. A suggested link is only valuable once it survives lab work and clinical trials, which is still where most of the time and cost of a new treatment goes.
Frequently Asked Questions
How were Toronto researchers using IBM Watson for Parkinson's disease?
The Ontario Brain Institute and University Health Network in Toronto adopted IBM Watson for Drug Discovery to analyze existing research with natural language processing and machine learning. The aim was to find overlooked connections, including compounds not previously considered for treating Parkinson's.
How much data could Watson for Drug Discovery draw on?
Watson Drug Discovery's resources included 31 million sources of data. Dr. Lorraine Kaila of UHN said the platform could find connections researchers might otherwise need weeks or months to uncover.
Why is developing a new drug so expensive?
According to PhRMA, research for a new treatment costs an average of $2.6 billion and takes 10 years. Only 12% of proposed medications make it past a Phase I clinical trial.
How common is Parkinson's disease in Ontario?
According to the Ontario Brain Institute, an average of 25 people are diagnosed with Parkinson's each day. An estimated 285,000 people in Ontario live with some form of neurodegenerative disorder.
What happened to IBM Watson Health after 2016?
A 2019 IEEE Spectrum investigation said IBM Watson had overpromised and underdelivered in health care. In January 2022, IBM agreed to sell Watson Health's data and analytics assets to Francisco Partners, and they became a standalone company called Merative.