IBM Watson Struggles To Deliver in Healthcare Segment

Watson Goes Live at Cleveland Clinic

There is great excitement around the role that IBM’s Watson supercomputer might one day play in healthcare. From data diving through EHRs to compile meaningful summary sheets, to helping doctors diagnose patients faster and more accurately by analyzing lab and exam findings and calculating probability-driven lists of likely diseases. For many, Watson is representative of the future of healthcare IT, a hybrid of heavy-weight buzz words like big data and personalized medicine.

For those that believe that supercomputers will one day support physicians and care delivery at this very sophisticated and intimate level, there is recent bad news on the Watson front. Primarily, Watson isn’t making IBM any money. Three years after IBM began trying to turn Watson into a profitable business venture, the project has a total revenue of less than $100 million. This is far from IBM’s forecast of $1 billion in Watson-generated revenue by 2018.

IBM has pursued the healthcare market the most aggressively because it believes healthcare is where Watson stands to have the most impact. With more than 17 percent of the US GDP spent on healthcare, this wasn’t a bad bet on IBM’s part. The problem is that for Watson to work, IBM engineers must first learn the technical processes in place that drive a customers business, and then they must program the supercomputer to begin analyzing this process. From here, Watson will begin to learn, and improve on its own over time. The process works, but it is slow and costly.

WellPoint, a health insurance company, is using Watson to evaluate the medical necessity of physician’s requested treatments and make approve/deny decisions. Initially, Watson was taking far too long to “learn” how to come to the correct decisions. IBM had to go back to the company and rework Watson’s learning analytics to get the project back on track. Since then, things have improved and WellPoint CEO Joseph Swedish says that Watson “has become a part of the way we do business.”

At M.D. Anderson Cancer Center, physicians started training Watson to evaluate leukemia patients two years ago. Watson’s objective was to help create personalized treatment plans based on outcomes probability. The project, according to Manoj Saxena, the executive overseeing Watson was "in a ditch" by early 2013. Fortunately, Watson is now getting close to the point that it might be able to assist in actual care planning for Leukemia cases. Lynda Chin, chair of genomic medicine at M.D. Anderson reports that Watson may be ready for service within 2014, but that it would be another two years before it could assist in any other types of cancer.

There is no doubt that Watson has potential. The problem is that it will take years, probably decades, for engineers to teach Watson how to correctly evaluate evidence and make sophisticated judgment-based decisions on enough different conditions to make it a must have commodity in healthcare. While many may argue that the potential is worth the wait, it may not be to IBM who is footing the bill for the time being.


Enjoy HIStalk Connect? Sign up for update alerts, or follow us at @HIStalkConnect.

↑ Back to top

Founding Sponsors

Platinum Sponsors