Elementary, My Dear Watson: IBM Enters Medical Image Storage
(A version of this Health Alert was published by Forbes on 8/17/2015.)
IBM has suffered declining revenues for 13 consecutive quarters. Although (like many U.S. companies) it attributes its poor results to the strong dollar, its once-praised pivot from hardware to software has put in into some crowded and stagnant markets. Nevertheless, it remains a company with great ambitions, largely built around Watson, the supercomputer which famously beat human beings on the game show Jeopardy! In 2011.
IBM has always hoped Watson would transform health care. It launched Watson Health as a strategic business unit in April, since which it has announced three small but significant acquisitions. Last week, IBM announced the friendly takeover of Merge Healthcare for $1 billion.
IBM bid $7.13 cash for the common stock on August 6, which caused it to rally from the previous day’s close of $5.41 to $7.10. This indicates that investors are highly confident the acquisition will close under the announced terms.
Merge is in the business of storing medical images, and IBM has big plans to transform that business. Although stores of medical images should be very valuable in this new era of cloud computing and big data, the industry has actually been undergoing some consolidation recently. Rumors that Merge was for sale go back at least to 2012; and Merge itself acquired the smaller DR Systems in February for $70 million. Merge could reasonably be categorized as the only pure-play medical image archiving business publicly listed in the U.S., and its stock has more than quintupled since the February 2009 market low, significantly greater than the rise in the NASDAQ 500 index.
This despite a Frost & Sullivan market research report published in October 2012, which concluded the worldwide medical image archiving market (worldwide) is set to double between 2013 and 2018. Frost & Sullivan listed the 15 leading companies in the space. Besides Merge, two of the fifteen have been acquired since the report was published.
Lexmark, a U.S.-based enterprise information management services provider bought privately held Acuo in 2013; and Japan’s Fujifim bought U.S.-based TeraMedica, Inc. (which spun out of the Mayo Clinic) last May. Other publicly listed firms with operations in this space are General Electric, McKesson, Philips, Siemens, Agfa-Gevaert of Belgium, and Sectra of Sweden. The leading privately held players are Dell, Carestream (which Onex, a Canadian private-equity sponsor, lifted out of the bankrupt Eastman Kodak) and three companies – Dejarnette Research Systems, Laitek, and Mach 7 – which are still controlled by their founders and close investors.
Some of these firms characterize health care as one vertical among many (for example, banking and retail) into which they sell their information management and image management solutions. IBM appears to be approaching the opportunity differently, maintaining a sharp division between Watson Health and other IBM services. According to IBM: “medical images are by far the largest and fastest-growing data source in the healthcare industry and perhaps the world – IBM researchers estimate that they account for at least 90 percent of all medical data today.”
Yet, according to IBM’s thesis, they are under-used. To date, Watson has excelled at natural language processing, which is useful for doctors’ notes and claims processing. However, medical image archives have not caught up. IBM promises that Watson’s artificial intelligence will “see” all the medical images in its archives and learn from them, bringing answers to radiologists and cardiologists that they cannot see with their naked eyes. The benefits of this approach to diagnosis, treatment, and research & development would be astonishing.
There are obstacles. One is the cockamamie U.S. health payment system. Although we are on the cusp of leveraging the information in medical images to an unprecedented degree, the bias in U.S. health care is that there is too much imaging. Medicare’s fees and use of medical imaging have been declining for over five years. Not that we should trivialize the risks of radiation or falsely positive diagnosis, but we do need to move beyond the fear of “over diagnosis” to valuing more detailed diagnosis – not to mention adding to the library of medical knowledge.
Another challenge is people’s legitimate concern over privacy of their data, as their medical images enter what is likely to be an increasingly accessible market. Like I wrote about genomic data in a recent article, the value of medical images is likely advancing faster than our federal and state privacy laws can accommodate.
I yield to optimism: The success of the 21st Century Cures legislation, which promises to revolutionize the U.S. regulatory environment for medical research, indicates that Congress retains the capacity to respond to the promise of innovation with responsible reform.
John–I loved seeing this column. You are right to focus on Watson for a bit as it has great promise. I agree with you that the over-use of imaging is a system-wide problem that will affect Watson’s impact if it continues and if there are heavy handed attempts to limit over-use.
But I’m writing for another reason. I heard the President of IBM interviewed at the Tech Museum in Mountain View a week ago. She was very impressivs as a person and as a visionary business leader. But Watson is aimed at the patient care event–a micro-use, you might say. I crave a Watson for the macro-level of healthcare–the point where purchasers, health plans and providers meet, and make decisions. This corral is no more sophisticated than it was in the Eighties. And, there is no resource that supports the simulation work that should be done. Moreover, each of those sectors is guided by a very ignorant Congress and bureaucracy, where old-fashioned ideas hit and often stick for ideological reasons, not because they work. One current example is the bundled payment experiment, now found not to work but which the Feds intend to mandate anyway.
As long as the Feds insist on micro-managing the healthcare system, the best defense is a truthful depiction of the probable effects of their rulings, on top of the changes that the sectors above have agreed to anyway.
The macro-system should not be micro-managed at the level of individual patient care events, period. But until now there hasn’t been even a hit of a promise that considering the system at the macro level was feasible. Watson may make it possible. There is another macro issue to consider: the effect of unfunded liabilities for Medicare and other government entitlements. The public needs to see and know what a total system cost picture would look like so it is prepared to accept adjustments to Social Security and Medicare.
Now, at this time there is one Presidential candidate that appears willing to carry the freight of a reform plan and that is Mark, the governor of Wisconsin, or is it Minnesota?
I’m thinking of asking the IBM lady to consider a macro application for Watson and to announce it during the campaign, as it will separate the sheep from the goats and give her a great PR opportunity, as well. I’m now outside of our 501 3 6 so can be as political as I want. Would you like to think this through with me?
I can outline what could be done and send it along.
Before you tell me, I can see a whole freeway of objections; privacy of business data, the sheer complexity of the decisions to be made, the fluid nature of healthcare so you can’t really pin down much of it via historical data, and the tendency of pundits and the press to pounce on anything negative and blow it out of proportions. All this is true and will not go away. I suppose it would be better to just let the intellectual mess fester. Can’t hurt to think about it, however.
Miss talking with you…..
Wanda J
San Francisco
Thank you. How would you handle privacy objections? It is a big hurdle.