Apologies on the hiatus for posting on THCB. As many of you know, I was running around getting Health 2.0 in order this past weekend. Today we are featuring a piece on understanding how machine learning can actually work in health care today-Matthew Holt
By LEONARD D’ AVOLIO, PhD
There’s plenty of coverage on what machine learning may do for healthcare and when. Painfully little has been written for non-technical healthcare leaders whose job it is to successfully execute in the real world with real returns. It’s time to address that gap for two reasons.
First, if you are responsible for improving care, operations, and/or the bottom line in a value-based environment, you will soon be forced to make decisions related to machine learning. Second, the way this stuff actually works is incredibly inconsistent with the way it’s being sold and the way we’re used to using data/information technology in healthcare.
I’ve been fortunate to have spent the past dozen years designing machine learning-powered solutions for healthcare across hundreds of academic medical centers, international public health projects, and health plans as a researcher, consultant, director, and CEO. Here’s a list of what I wish I had known years ago.
This week I attended an all day “training” session in a new medical record system. I thought it was interesting that the experience was called “training,” which prompted me to remind myself of a few useful definitions.
Education, from the Latin root meaning a drawing forth, implies not so much the communication of knowledge as the discipline of the intellect; an intra-cerebral process aimed in large part at creating principles upon which new knowledge may be elaborated. Instruction is that part of education that furnishes the mind with knowledge. Teaching is often applied to practice as in “teaching a dog to do tricks.”
Training is an element of education in which the chief characteristic is exercise or practice for the purpose of imparting facility, as in “training for the marathon.” Breeding relates to manners and outward conduct as in “standing when elders enter a room is a sign of good breeding.” Regimentation is the prescription of a particular way of life or thinking usually involving the imposition of discipline. The term, arising from military regiment, is related to the medical usage of regimen, as in “the patient keeps his prescription medications in separate compartments of a plastic container in order to accurately adhere to his regimen.” Propaganda is the systematic propagation of a doctrine, cause or information reflecting the views and interests of those advocating such a doctrine or cause, as in “ACCME is propagating the view that elaborate re-certification maneuvers will improve the lives of patients.”
A cheerful instructor started the session by asking each of us to introduce ourselves and reveal a “secret guilty pleasure.” Mine is to create elaborate cocktails. If only I had had one of my famous Marty’s Beerjitos with me the whole experience could have been much more pleasant. In addition to the instructor, there were several “super-users” in the room to facilitate the process. It was immediately obvious to me that the super-users hovered behind my chair. These friendly young people had correctly identified me a “super-loser.” Had I been litigious I would have reported the experience to our ombudsperson as blatant ageism.
But, alas, they were correct. I was hopeless. Besides, I don’t believe in ombudspeople. I believe one should speak for oneself.
Our friends at Kaiser Permanente asked us reach out THCB readers for help with a cool crowdsourcing project. The Kaiser innovation team is working on developing new content for The Kaiser Permanente Center For Total Health , KP’s shiny new 16,000 square foot exhibition and meeting space in downtown Washington D.C.
If you’re close enough to make the trip, we highly recommend that you stop by and take an hour or so to poke around a bit before submitting your suggestions. Failing that, you can take the online interactive tour here.
If you’re a doctor, a med student, a designer, an entrepreneur, a patient – or if you just have a good idea – we’d like to hear from you. KP’s innovation team asked us to ask you four questions. You can answer one or you can answer them all.
1. What is Total Health? In other words, what is health? What’s important to you?
2. What should total health look like when implemented? What innovations can be used to drive change in the healthcare system? What will healthcare look like in the future?
3.How should total health be supported? What can be done to make healthcare better? Smarter? Both within the healthcare system? And in our own lives?
4. If you were designing an interactive wall to demonstrate total health to visitors what would you focus on. In other words, if you were designing an exhibition what would it look like? What would your message be? What would help educate the public? How would you get that message across? Yes, you can send us an picture.
Answers can be left in the comment thread below. If you prefer to submit a video response via YouTube send the link to email@example.com. or paste in the comments below. Blog posts should be submitted to THCB editors at firstname.lastname@example.org
For the interactive design question, we asked THCB’s editors what they’d like to see. Here’s what we came up with on the back of our paper napkin:
Coursera, the popular massive open online course (MOOC) platform, intrigues. With over 5 million students served and $85 million raised—both numbers are first among the “MOOC platforms”—it’s the type of company that captures the imagination of people in Silicon Valley who dream of transforming sectors.
But Coursera has always given me reason to pause as well. It’s never felt to me like its initial incarnation could possibly disrupt higher education. Why? As I’ve told its team, offering courses from the top universities online and claiming that at last, anyone anywhere can access the best learning in the world isn’t correct.
The reason is that the top universities do not offer the best teaching and learning experiences. Instead, their faculty members are incentivized heavily to focus on research at the expense of teaching. If a professor seeking tenure at one of these institutions receives a teaching award, it is often said that that professor has just received the kiss of death for her tenure hopes. If students learn at these institutions, it’s often not because the teaching is so good, but because the students are so talented that they can absorb anything thrown at them (and it’s worth noting that just because a professor is entertaining, does not mean it’s a good learning experience).
Putting these courses online often makes them worse. Not only do professors not know how to teach well in person, but also their lack of understanding of the basic principles of sound learning design causes them to exacerbate these problems as they put these experiences online, which can become more problematic as students from all walks of life with many different learning needs are now theoretically able to take these courses.
I remember when one of my patients with coronary artery disease suggested that he be given a course of an antibiotic to lower his future risk of a heart attack. The patient had done his homework, quoting literature that pointed to a possible infectious link to atherosclerosis. He also was aware of the theory that aspirin’s benefit had less to do with blood thinning than reducing underlying inflammation.
All of which makes me wonder if the vaunted Doctor of Medicine degree may be vulnerable.
Why should physician education be immune from a perfect storm of over-priced graduate education, “alternative” web-enabled learning with on-the-job-training? The declining value of the formal credential may be less about the university degree and more about competency, turbocharged by flexible licensing and a discerning consumer.
Non-physician health care professionals are arguing that their expertise is enough to enable them to deliver babies, administer anesthesia, prescribe drugs and perform surgery. My traditionalist colleagues argue that patient safety is at stake and that lay persons may not be able to discern all of the possible risks, benefits and alternatives. When things go occasionally wrong in the delivery suit, operating room or with a drug, they say a credentialed and experienced doc can make the difference between life and death.
If Americans judged the quality of hospital care the way Newsweek judges high schools, we would soon be inundated with “charter hospitals” that only treat healthy patients.
As reported in The New York Times, thirty-seven of Newsweek’s top 50 high schools have selective admission standards, thereby enrolling the cream of the eighth grade crop. That means that when these high scoring eighth graders reach eleventh grade, they’ll be high scoring eleventh graders, helping the school move up the Newsweek rankings. These selective admission schools simply have to avoid screwing up their talented students.
That’s no way to determine how good a school is. The measure of a good education should be to assess how well students did in that school compared to how they would have been predicted to do if they had gone to other schools.
Imagine two liver transplant programs, one whose patients experience 90% survival in the year following their transplant and the other whose patients experience only a 75% survival rate. Based on that information, the former hospital looks like the place to go when your liver fails. But aren’t you curious about the kind of patients that receive care in these two hospitals? Wouldn’t you want to know whether that first hospital was padding its statistics by selectively transplanting relatively healthy patients?
Nearly a year ago, one of my blog posts bemoaned a gap in our training of future physicians—a lack of training in the skills needed to lead projects in patient safety and quality improvement.
I wrote the post after speaking to a group of medical students who were energized about this area of work. Yet, as I reflected on the talk:
“I had to confront the sad reality that most of them will graduate ill-prepared to lead the improvements of quality and safety our health care system needs. They no doubt will know chemistry, biology and physiology, but they may not know about human factors, implementation science or performance measurement—the language of quality improvement. They will know orthopedics and genetics but they won’t know teamwork and systems engineering. They likely know about German scientist Rudolph Virchow, the father of cell theory, yet they do not know John Kotter, the father of change theory whose model for leading change is highly effective and widely used.”
So how can medical students, residents and fellows make quality improvement and patient safety a focus of their clinical careers? On Nov. 10, the Armstrong Institute and the American College of Medical Quality will be hosting the National Workshop on Quality for Medical Education—affordable and open to anyone—that focuses on how medical students, residents and fellows can integrate safety and quality into their clinical careers. What career paths exist? What tools and skills are needed to carry out this work, and where do you get them? What kinds of quality and safety projects are residents and students taking on?
There’s a larger question here about why the scholarly world allows itself to be judged by secretive Scandinavian committees sitting on endowments funded by money made selling explosives. But let’s put anti-Nobel polemics aside.
The announcement today that Alvin Roth and Lloyd Shapley won this year’s award in economics came with the explanation that they had devised systems for matching buyers and sellers that led to more rational outcomes than existing markets.
Shapley, a contemporary of “A Beautiful Mind’s” John Nash, introduced an elegant theory 50 years ago to explain the (relative) stability of marriage pairings despite the fact that individuals have complicated preferences when choosing a mate. Shapley’s idea is that the person you end up with is the best match given everyone else’s preferences.
You might prefer someone else more than your current mate, but that person has you lower on her list, and so on. Imagine Larry. If he could Larry would have definitely married Elizabeth Taylor. But she was taken so now Larry is happy with his actual wife. (To sum it up in a way that would make an economist cringe.)
Alvin Roth built on that early theory. He designed actual markets that used the matching principle, also known as the deferred acceptance algorithm, as a guiding principle. The most famous example of a Roth market is the Residency Match.
Medical residency is a job that lasts three to seven years, depending on the program, and follows graduation from medical school. It is required for a doctor to complete a residency in order to be licensed to practice.
In the “old days” medical students would apply to hospitals and rank their preferences in a way that was visible to those institutions. A hospital would first review those applicants who had indicated them as the first choice. If spots remained to fill, a hospital would then look at those who had picked it second, and so on. You can quickly see how this system punished people who shot high and missed. They would end up at one of their last choices because the best places would fill up quickly.
I went to medical school in Cleveland and did myj pathology residency in San Francisco at UCSF. I was on the medical school faculty at UCSF, Iowa, Allegheny University of the Health Sciences, and Michigan State.
Since leaving academic medicine, I have worked at a bio-tech start up in Cambridge, an educational and research institute in Grand Rapids, a $2 billion integrated delivery system in Iowa, and an evidence-based medicine consortium in Minneapolis.
In my experience physician executive positions do not always last a long time because the environment changes, my career aspirations changed, and getting the job done sometimes means alienating enough people to get in the way of long job tenure.
2. You Will Have to Reinvent Yourself Over and Over Again
My main professional roles have included: medical school pathology course master, surgical pathologist, division head, vice chair of academic department, chair of academic department, medical director of managed care, corporate operations officer of ambulatory care, special assistant to the president of a big ten university for managed care, search consultant, chief knowledge officer of a genomics bio-tech start up, president and ceo of an educational consortium, chief medical officer of a delivery system, president and ceo of an evidence based medicine institute, and health policy professor at a school of population health.