As I walk into the building, the sheer grandiosity of the room is one to withhold — it’s as if I’m walking into Grand Central station. There’s a small army of people, all busy at their desks, working to carry out the next wave of innovations helping more than a million lives within the Greater Philadelphia region. However, I’m not here to catch a train or enjoy the sights. I’m at the office of the President and CEO of Thomas Jefferson University, Dr. Stephen Klasko, currently at the helm of one of the largest healthcare systems in the U.S.
Let me backup a little.
The theme of nearly every conversation about the future of technology now revolves around Artificial Intelligence (AI). Much weight is placed on the potential capacity of AI to disrupt industries and change them to the very core. This pressure has been felt to a large extent within nearly every aspect of healthcare where AI has been projected to improve patient care delivery while saving billions of dollars.
Unfortunately, most discussions exploring the implications of AI only superficially look at either the product or the algorithm that powers these products. The short-sightedness of this approach is not an easy one to fix. Yes, clinical studies validating AI backed products are vital but AI cannot be viewed just like any other drug or a medical device. There’s much more to be considered when we examine the broader role of this technology, because this technology can shape the entire healthcare system. To place the impact of a far reaching technology, you need an even longer sighted vision. It’s a rare breed of people that have experienced the tumultuous history of change within medicine but can still call upon the lessons learned to execute innovations and bring meaningful results.
For more than a decade, a running joke among electronic health (EHR) record skeptics has been that its clunky “decision support” functions, defined as the on-screen provision of clinical knowledge and patient information that helps physicians enhance patient care , is condemned to always remain an innovation of the future. Yet, while published studies like this continue to fuel doubt about the prime-time readiness of this EHR-based technology, a growing body of clinical research suggests that the science is getting better. Jonathan Cohn, writing in The Atlantic, points out that IBM’s Watson has achieved enough of a level of sophistication to warrant clinical trials at prestigious institutions such as the Cleveland Clinic and Memorial Sloan-Kettering.
Unfortunately, there is an under-recognized threat to EHR-based decision support: the dysfunctional U.S. tort system.
The experience of Google’s “driverless car” may be instructive. According to National Public Radio, years of testing is putting this technology within reach of consumers. Thanks to the prospect of fewer accidents, better transportation options for the disabled, reduced traffic congestion and lower hydrocarbon consumption, some states have responded by attempting to support this promising technology with “enabling legislation.”
Unfortunately, the legislation in some state jurisdictions is being hindered by the prospect of complicated lawsuits. As physicians know all too well, when a single mishap lands in court, adroit attorneys can use the legal doctrine of joint and several liability to tap multiple deep pockets to increase the potential size of the award. In the case of driverless cars – in which the owner is more of a passenger than a driver – the accidents that are bound to happen could metastasize upstream from the owner and tie up the driverless automobile manufacturers and all of their business partners in time-consuming and expensive litigation.
Ditto the EHR’s decision support technology. Even with Watson’s intelligence, medicine will remain imperfect and allegations of medical mistakes will be inevitable. When lawsuits arise, the defendant medical providers will likely argue that their judgment was clouded by the very technology that otherwise helped them better serve their other patients. Personal injury lawyers are unlikely to let that theory of liability go unused. Tapping the same kind of lucrative joint-and-several legal theories that have served them so well in decades of standard malpractice litigation, they’ll undoubtedly be happy to name the EHR manufacturer and all of its decision-support business partners in these lawsuits.
One of the computer applications that has received the most attention in healthcare is Watson, the IBM system that achieved fame by beating humans at the television game show, Jeopardy!. Sometimes it seems there is such hype around Watson that people do not realize what the system actually does. Watson is a type of computer application known as a “question-answering system.” It works similarly to a search engine, but instead of retrieving “documents” (e.g., articles, Web pages, images, etc.), it outputs “answers” (or at least short snippets of text that are likely to contain answers to questions posed to it).
As one who has done research in information retrieval (IR, also sometimes called “search”) for over two decades, I am interested in how Watson works and how well it performs on the tasks for which it is used. As someone also interested in IR applied to health and biomedicine, I am even more curious about its healthcare applications. Since winning at Jeopardy!, Watson has “graduated medical school” and “started its medical career”. The latter reference touts Watson as an alternative to the “meaningful use” program providing incentives for electronic health record (EHR) adoption, but I see Watson as a very different application, and one potentially benefitting from the growing quantity of clinical data, especially the standards-based data we will hopefully see in Stage 2 of the program. (I also have skepticism for some of these proposed uses of Watson, such as its “crunching” through EHR data to “learn” medicine. Those advocating Watson performing this task need to understand the limits to observational studies in medicine.)
One concern I have had about Watson is that the publicity around it has been mostly news articles and press releases. As an evidence-based informatician, I would like to see more scientific analysis, i.e., what does Watson do to improve healthcare and how successful is it at doing so? I was therefore pleased to come across a journal article evaluating Watson . In this first evaluation in the medical domain, Watson was trained using several resources from internal medicine, such as ACP Medicine, PIER, Merck Manual, and MKSAP. Watson was applied, and further trained with 5000 questions, in Doctor’s Dilemma, a competition somewhat like Jeopardy! that is run by American College of Physicians and in which medical trainees participate each year. A sample question from the paper is, Familial adenomatous polyposis is caused by mutations of this gene, with the answer being, APC Gene. (Googling the text of the question gives the correct answer at the top of its ranking to this and the two other sample questions provided in the paper).
Watson was evaluated on an additional 188 unseen questions . The primary outcome measure was recall (number of correct answers) at 10 results shown, and performance varied from 0.49 for the baseline system to 0.77 for the fully adapted and trained system. In other words, looking at the top ten answers for these 188 questions, 77% of those Watson provided were correct.
This past week, Google had its annual developers conference, Google I/O. One of the more provocative talks, called “The End of Search as We Know It,” was by Amit Singhal, who is in charge of search for Google.
The vision, as described by Amit, is that instead of typing words into a box on a website or mobile app, we will have conversations with Google, enabling a much more personalized, refined experience. The holy grail, of course, is that Google analytics become both predictive and prescriptive, serving you content that is just right for you and anticipates your needs.
It seems there is a race on now to achieve this vision. One could argue that Amazon, Apple, Facebook, Pandora and others are all in the same mode. Best I can tell, the promise these companies are floating to advertisers is that their ads will be served up to that focused slice of the population that will find their product relevant in the moment.
If you apply this thinking to healthcare, several controversies/topics come to the fore.
Is Google competing with IBM’s Watson? Undoubtedly yes. On the other hand, I’m guessing Google is disenchanted with the consumer health space after the demise of its personal health record (PHR). And IBM seems to be focused on clinician decision support. So early in the game, with respect to healthcare anyway, maybe there is not much competition. The path for clinician decision support is clear and the market obvious, whereas the path and market for consumer health decision support are blurry.
Little over a month ago, IBM and WellPoint announced an agreement wherein WellPoint will deploy IBM’s latest and greatest super computer and artificial intelligence mega-mind Watson. Watson’s claim to fame was its ability to beat the human Jeopardy champions much like Big Blue beat reigning chess champion Garry Kasparov in 1997. Since that Jeopardy match, IBM has been quite vocal about its desire to apply Watson in the medical arena, we’ve been buried in press releases and briefings, but the WellPoint announcement is the first one of any real consequence. Having interviewed both IBM and WellPoint, following is our review and assessment.
Watson is a relatively new form of artificial intelligence, based to some extent on neural networks. What is unique about Watson is that it has been developed (trained) to understand the nuances of language. It is a question & answer system that uses among other techniques, natural language processing, to extract meaning out of unstructured data. In developing Watson for the Jeopardy challenge, one of the key design parameters was for Watson to answer a question in under three seconds – plenty fast enough in a diagnosis/treatment decision scenario. This is a key reason why Watson may have enormous utility in the healthcare sector where so much data is unstructured, the pace of change is so high and the ability to chose the optimum treatment patient plan for a given diagnosis is less than ideal today.
IBM’s Jeopardy-champion computer, Watson, has huge potential for helping physicians and other clinicians work with patients.
The leap from TV game show to physicians’ offices will probably take at least two years. But Watson’s understanding of natural language, vast storehouse of information and ability to keep up with rapidly changing medical research could significantly improve medical care.
The medical faculty at Columbia University and University of Maryland are helping program a Watson-type computer to assist clinicians.
A few years from now, consulting Watson could become a routine part of a clinician’s practice. Caregivers have traditionally resisted computerized assistance in diagnosis and treatment because the technology has been awkward to use and questionnaire-based systems have been too rigid. But Watson can “understand” descriptions of a patient’s symptoms in natural language, and it can even scan years of medical records and doctors’ notes to determine what diagnostic and therapeutic options it might suggest. Doctors can ask it questions using the same terms they would use in an e-mail to a colleague. Continue reading…
Last week we all watched in awe as the IBM computer, Watson, trounced two of Jeopardy’s finest. This event has been much heralded but it is worth stopping for just a minute to reflect on the experience of watching Jeopardy those three nights. I had no trouble rooting for Watson, feeling disappointed or embarrassed when he missed a question and chuckling when he displayed any behavior that seemed the least bit human. I knew the whole time, on one level, that Watson is a computer. On another level though, I bonded with him and felt a good deal of emotion regarding his success.
MIT Prof. Sherry Turkle recently released a book entitled Alone Together. She was also interviewed recently on TechCrunch. Turkle puts forth the view that technology is a poor substitute for interaction with a human being. However, she notes that when technologies (robots, relational agents and the like) respond to us, they push “Darwinian buttons,” prompting us to create a mental construct that we are interacting with a sentient being. This brings a host of emotions to the communication including affection. Turkle makes an argument that in the realm of human relationships this phenomenon is unhealthy for our species.
I’d like to bring in principles from behavioral psychologist, Robert Cialdini, who has authored several books on the psychology of persuasion. Cialdini offers simple tools that can be used in everyday life to persuade others to adopt one’s point of view. In doing so, he lays out solid experimental evidence that these tools are effective, in most cases without the recipient being aware. Continue reading…
Game Show Watson wants to be a doctor. Well, almost.
Fresh off a commanding victory on Jeopardy, IBM will try to demonstrate that the combination of advanced natural language processing and sophisticated algorithmic decision-making capabilities involved in its extraordinary Watson computer can help humankind, not merely humiliate human competitors.
As I wrote on a previous blog, IBM began eying the medical marketplace more than 45 years ago. IBM CEO Thomas J. Watson, Jr. – son of the IBM CEO for whom this computer was named – put it this way in 1965: “The widespread use [of computers]…in hospitals and physicians’ offices will instantaneously give a doctor or a nurse a patient’s entire medical history, eliminating both guesswork and bad recollection, and sometimes making a difference between life and death.”
Now, IBM is ready to turn that vision into reality. At heart, Watson is the world’s most sophisticated question-answering machine. The company is collaborating with Columbia University and the University of Maryland to create a physician’s assistant service that will allow doctors to query a cybernetic assistant. IBM will also work with Nuance Communications, Inc. to add voice recognition to the physician’s assistant, “possibly making the service available in as little as 18 months.” For Nuance, it could be a major business line, and promises to carry over in the not too distant future to the mobile phone market, such as Apple’s iPhone, where Nuance is a major presence.Continue reading…
If you want to see the future of health information technology, take a look at the dueling visions of two Thomas Watsons that are on display this month in a game show and a trade show. The juxtaposition unintentionally demonstrates what doctors and patients will be doing together and also what they can do separately.
What I’ll call Game Show Watson is a computer named for IBM founder Thomas J. Watson, Sr. This Watson is appearing on the TV show Jeopardy to play a highly publicized set of matches against two human champions from Feb. 14-16. Although viewers will actually see a black computer screen with a revolving blue globe, Game Show Watson itself, in the tradition of “Big Iron” mainframes, consists of ten refrigerator-sized servers located offstage.
In contrast, the Watson at the trade show is not one computer, but thousands of them, all contained inside the mobile devices that are descendants of the telephone first demonstrated by Alexander Graham Bell and his assistant, Thomas A. Watson. (That Watson was also an inventor is a topic for another time.) The Telephone Watsons, on display for the tens of thousands of attendees at HIMSS11 from Feb. 20-24, are giving rise to a new field known as “mobile health.”Continue reading…