Recently, there have been some articles about IBM Watson and healthcare. In particular, a blog by John Lynn speculates on the “health” of IBM Watson in healthcare.
I’ve had my own personal experiences with IBM in healthcare. While with GE, we did a joint venture with IBM in the early days of PACS and RIS, and I was an original IBM MedSpeak/Radiology reseller in the early days of speech recognition. In both cases, IBM did not have the financial wherewithal to play the long game in terms of the technology and exited both.
Interestingly enough, MedSpeak/Radiology was not the first foray into digital dictation for IBM, so they exited that business multiple times. I also have to say, MedSpeak/Radiology was a pioneering technology that was very competitive. It just wasn’t the business case the IBM management was looking for. Similarly, IBM’s effort with GE toward RIS and PACS was naively predicated on employing IBM’s RIS product, and not on interoperability with other RIS vendors.
There are countless business examples of companies launching products based on great expectations, only to have them fail to achieve those expectations. There are also countless examples of companies getting into products that have no relevance to existing businesses because the think they have a “better mousetrap.” One that comes to mind is GE. Years ago, GE developed in their research center an electric garden tractor which was very innovative for the 1970’s. Unfortunately, when GE attempted to bring the product to market, they discovered they did not have a single market channel that addressed lawn-care equipment! GE ended up selling it to Bolens for a fraction of the development cost. And, in retrospect, it was premature given the current fascination with electric vehicles!
I am not implying that IBM hasn’t done its homework on IBM Watson. I suspect it may be more of two other factors at play. One, it’s a question of chicken and egg. Is the advancement of Artificial Intelligence more about the application or the platform? Secondly, does IBM have the infrastructure to succeed? In the case of the first factor, IBM is more about the platform than the application. In the second case, the same could be said for the platform versus the application, as IBM does not have a significantly parallel channel that would address AI applications.
I think this is why IBM initially acquired Merge Technologies, as they saw it as both a “sandbox” to learn imaging, but they may have also seen it as a potential distribution channel. With further understanding, they were quick to learn that attempting to develop and market applications through Merge would make them a competitor to other viable players such as GE, Philips, and Siemens. That resulted in a move to create “consortiums” that could develop applications on the IBM Watson platform, thus broadening the appeal of IBM Watson across a broader distribution channel.
The question going forward is – has IBM learned from its history of past healthcare ventures? If IBM can make a business case for addressing Watson as the platform for AI, it might be a stronger case than trying to be the end point for AI. This would be consistent with its strength in computational platforms in business, which have classically been a better business model than applications (OS2 or Lotus vs. Microsoft Windows and Office anybody?!).
I recall the OS2 days versus Microsoft Windows. As a fan of OS2, I had hoped it would succeed. Unfortunately, as I learned, there were very few applications for OS2, which is ultimately why people purchased personal computers in the first place. If IBM had been smart, they would have given OS2 software licenses away in an attempt to build consumer demand for the development of applications. Unfortunately, that didn’t happen, and I doubt there are any OS2 users out there today.
If IBM can take stock of its history, there is a fighting chance for IBM Watson to be a significant factor in healthcare. But, the company will need to learn how best to foster the development of AI applications. In his blog, John Lynn relates the experience of Lukas Wartman of the McDonnell Genome Institute at the Washington University School of Medicine in St. Louis, who laments on how much faith one can put in the results (of using Watson). As with past technologies, we are at the tip of the iceberg in terms of applying the technology. Here’s hoping that IBM can figure this one out, be in it for the long game, and live up to the hype of IBM Watson!
