The general public might look at the Armonk, N.Y.-based International Business Machines Corporation, universally known as IBM, and still think of the multinational organization as a computer hardware and software company only. However, few areas of health and medicine have gone untouched by the technology, research and innovation generated by IBM in recent decades.
Having originated more than 100 years ago, the company has continuously evolved since its inception. Over the last several years specifically, IBM has been focused on helping its provider organization clients in their objectives to serve their populations, and be more patient-centered and outcome-oriented in the work they do, says Sean Hogan, vice president and general manager of healthcare at IBM. “Now that we have had dramatic acceleration and adoption of electronic medical record (EMR) systems and other sources of information, how do we help organizations take advantage of that as an asset to better perform?” Hogan says, noting that IBM “doesn’t define itself as a product company, but one that innovates around solving problems.”
At the core of that, he adds, is doing analytics work for organizations, helping them get access to data and get confidence in their data, as well as completeness of it. The organization’s data analytics software already figures in prominent medical research trials with the Cleveland Clinic, Mayo Clinic, and the Memorial Sloan Kettering Cancer Center. “If you’re not confident about the quality and completeness of information, it’s difficult to get teams to execute against that. Helping organizations manage their information is a major driver for how they’re serving populations,” Hogan says.
Another big part of IBM’s mission, Hogan says, is helping organizations take advantage of its migrate-to-cloud-based services. As such, the company made two major moves in April at the Healthcare Information and Management Systems Society (HIMSS) conference in Chicago, announcing the acquisition of the Cleveland-based Explorys, a healthcare intelligence cloud firm, and Dallas-based population health management company Phytel. Regarding the Explorys acquisition, IBM officials noted that, “Since its spin-off from the Cleveland Clinic in 2009, Explorys has secured a robust healthcare database derived from numerous and diverse financial, operational and medical record systems comprising 315 billion longitudinal data points across the continuum of care.”
According to Anil Jain, M.D., chief medical officer (CMO) for Explorys, and practicing physician at the Cleveland Clinic, there is “clear synergy” among the two companies, IBM and Explorys. “Over the years we [Explorys] have become leaders when it comes to aggregating data from all the different disparate data sources that exist in a health system, and bringing it together to solve real-world problems in a very rapid manner using population health as a use case that is near and dear to most clinical systems right now,” Dr. Jain says. “In many ways it’s very much a competitive advantage to get a handle on all that data and do actionable things.” Jain says that what attracted IBM was Explorys’ ability to help them accelerate, from a data point of view—with its 50 million lives and 360 hospitals—but also from an analytics perspective.
The Era of Cognitive Computing
Enter the Watson Health Cloud, which IBM will sell to doctors, hospitals, insurers and patients. That offering will be the centerpiece of a new dedicated, Boston-area business unit, IBM Watson Health, which now includes both Explorys and Phytel. “[IBM] is complimenting much of what we do around traditional analytics using machine learning algorithms with some of the cognitive computing and the Watson analytics that Watson Health group will be leveraging,” Jain says. “We became the content that will fuel some of the next generation analytics that Watson has become famous for.”
Indeed, Watson, the Jeopardy!-playing supercomputer, is built to mirror the human learning process through the power of cognition. According to Rob Merkel, vice president of the Watson Health unit, the creation of this group “is a signal to the market that we are making significant advancements in the market to accelerate this technology.” At the time of the April announcement, Merkel says, it was about three years since the Jeopardy! demonstration. In that time span, the Watson team went from 26 researchers to 150 to ending last year with 2,000, he notes. “At the end of the day, we’re trying to provide insights that are beyond human cognition,” Merkel says.
Merkel says that the organization has been trying to accomplish this coming from two different directions—published knowledge and data-driven insights. Within healthcare, there’s a few data points that just make it clear what’s beyond cognition, he says. “I see all of these stats on the proliferation of medical literature, from it doubling every five years to every three years, and sometimes studies even say it will just takes days for medical literature to double,” he says. “There are 700,000 research articles and 180,000 clinical trials per year. Even if all you did was read, you wouldn’t be able to scratch the surface in many lifetimes. So that’s the knowledge approach—providing insights off those large knowledge repositories,” Merkel says.
The other side is data-driven insights—what is the information you can find out about an individual? According to Merkel, IBM research estimates that over the course of a lifetime, one person will generate about 400 gigabytes worth of information, a number that increases to 6 terabytes of information when genomics are factored in. Factoring in exogenous behavioral type data such as Fitbit information it goes to 1,100 terabytes, or as Merkel says, “11,000 top-of-the-line smartphones worth of information.” He adds, “There is just so much information to analyze, both knowledge-driven and data-driven. That’s where we’re coming from when we announce Watson health and the formation of the industry cloud that we will create to put in a standards-based, compliant, massively scalable information base to provide these insights off of,” he says.
To this end, Watson applications in healthcare are organized across three dimensions, Merkel says: consumer engagement, R&D related productivity solutions, and improving clinical outcomes. In the first category, consumer engagement, is IBM’s Engagement Advisor solution, which is what most people think of when they think of Jeopardy! demonstration. You ask a difficult question, you get a very good answer, Merkel notes. “Now it has advanced four years where you can have dialogue with Watson, and it remembers the context in which you’re having this dialogue,” he says. “We have done a lot of work around wellness solutions and chronic condition management solutions where a person can ask questions about his or her condition and have Watson give thoughtful answers,” he says.
In R&D transformation, the company’s Discovery Advisor solution is geared towards research in which hypotheses are generated and tested from Watson reading through 40 million research documents related to healthcare and life sciences. For example, says Merkel, at the Baylor College of Medicine, researchers were interested in phosphorylation, the p53 protein that stops the formation of tumors. There are some 70,000 studies in Medline alone on this single protein, Merkel says. “Now imagine Watson doing this for you as you’re trying to identify new targets for research. The entire run rate for the industry over the past 10 to 20 years has been approximately one [new target] per year. In weeks of using Watson, they identified seven which were peer reviewed and published,” Merkel says. Another IBM solution used in oncology allows Watson to receive information on a tumor which it then explores all known literature about, Merkel says. “Imagine [Watson] creates a complete biochemical picture of the tumor, which would take months worth of effort of highly-skilled people. Now imagine Watson could do that in five minutes. That’s what’s being beta tested today with this solution,” Merkel says.
In terms of clinical outcomes, Merkel points to a solution that IBM developed with Sloan Kettering which allows Watson to provide treatment recommendations, starting with chemotherapy but eventually expanding into other areas of oncology. Then there is its EMR Advisor, in which Watson reads through a complete patient longitudinal history where it can then assist a physician in understanding what’s going on in a patient’s history. As a novel example, Merkel notes, Watson dynamically generates the problem list based on reading through the clinician’s notes. “Imagine Watson creates a problem list based on what it sees in the medical record, something that physicians and nurses don’t have a lot of time to maintain, so it’s not often the most accurate. Side-by-side it has performed very well against humans,” he says.
Also in the EMR arena, IBM and the Verona, Wis.-based Epic Systems are collaborating with Mayo Clinic to bring EMRs together with the cognitive computing capabilities of Watson. They are working on a proof of concept focused on helping match patients to the most relevant clinical trials for their individual conditions. Epic is extracting patient data from health records, delivering it to Watson to be quickly compared with massive volumes of relevant clinical data, and then sending results back into the Epic EMR. For patients, this potentially means more rapid and thorough analysis of all the factors that could impact their care drawing from insights far beyond the information contained in medical records alone, say IBM officials.
All of this moving and shaking, according to Hogan and Merkel, presents an opportunity to embark into the “third era of computing.” First, says Merkel, is tabulating systems; second is programmable systems, and “now we are in the era of cognitive computing. We think this will last for decades and this information revolution will be enormous,” he says. Hogan agrees with the notion that healthcare is such an information-intensive industry, noting that it will require a major shift in terms of being able to unlock a better understanding of approaches to handling medical conditions and health statuses. “We have the ability to do advanced analytics to see patterns and surface outliers, added with the cognitive capability that brings forward context,” Hogan says. “So to be able to access medical knowledge and journals and literature, combine that with data specific to a person, and then apply all of the available medical knowledge against what’s known about someone and their condition—that could evolve knowledge,” he says.