Wednesday, May 31, 2017

Building a Better Health Care System

Several Kaiser Permanente executives recently urged that we stop wasting patients' time.  They cited the ultimately tragic example of Jess Jacobs, who used her Six Sigma expertise to calculate that only 4.75% of her outpatient visits and 0.08% of her inpatient stays were spent actually treating her conditions.  They then outlined how they've re-engineered some of Kaiser's processes "to upend traditional paradigms and make saving our patients’ time a part of our standard quality measures."

That's a start, but, if we're going to re-engineer health care processes, let's really re-engineer them.  The future is patient-driven, on-demand, and just-in-time.

Hey: it's the 21st century.  We have computers and smartphones, we have unobtrusive sensors, and we're beginning to have meaningful artificial intelligence (AI).  The health care system shouldn't still be built around waiting or even on in-person contact.  It may still include those, but they should be the exception, not the norm.

Let's look at what it might look like:

  • You wake up in the morning and will already have more information about your health than your doctor does now, between, say, biosensors on your skin, ingestible sensors,  even smart toilets.  
  • All that data will go to your personal health assistant, such as Baidu's Melody or  It will know your norms, understand when something appears off, and listen to your (health) complaints.  It will give you a good idea what might be wrong and what you should do about it.  Perhaps most importantly, it will tell you when you need to consult which doctor.
  • The first stop will usually be to consult an AI "doctor."  It be available 24/7, and will have access to all your data and all the deep learning only AIs can accumulate.  
  • If you do see a human doctor, chances are it will be first via a screen (or VR/AR).  They will be the best available expert for your problem; not just locally, but in the world.  In-person visits will be reserved for the most urgent problems requiring hands-on care.
  • When you do have to visit an office or facility, it won't be like today, full of forms, waiting, and uncertainty.  It will be more like visiting an expensive spa, where you are treated like royalty.  They'll be ready for you, cater to your unique needs, and get you in and out efficiently.  
  • "Minimally invasive" will take on whole new meanings, from ingestible robot surgeons to reprogramming our immune cells to tweaking our microbiome.  
  • With so much of our care AI-driven or AI-supported, there will be much less reliance on hunches or outdated research, and more on data and evidence of efficacy, reducing the current "epidemic" of over-testing and use of unnecessary procedures. 

Sound optimistic?  Ray Dorsey and Eric Topel write about the future of office visits, and while they don't go as far as the above, they do assert: "Tomorrow’s office visit will increasingly take place everywhere but the office."
They include on-demand home visits and hospital-at-home models in their prediction, pointing out:
While most patients like their doctor, almost no one likes going to the doctor. By contrast, tomorrow’s office visit will offer patients unprecedented access to confidential, expert care that is delivered conveniently in the comfort of their homes.
The big question is less where or how visits happen, but who is making them.  I.e., whether AI can truly supplant the role of human practitioners for much of our care.

Newsweek recently proclaimed that AI will "cure" our health care system, and claimed, "we’re close to being able to build AI software that can at least supplant that first visit to a doctor when you’re sick—which, of course, is when you least want to travel to a doctor’s office."

Their optimistic prediction:
Add it up, and in these next few years we’re going to see a parade of tech applications that reduce demand on the health care system while giving all of us more access to care. Doctors should be freed up to do a better job for patients who truly need their attention. 
CNBC's Christina Farr, interviewed several doctors about AI's future.  She found that physicians were skeptical of sweeping changes -- but many agreed they hoped AI could free them of more mundane tasks, which would help them spend more time with patients.

Rasu Shrestha, chief innovation officer for UPMC, told Ms. Farr: "Health and health care is too human a notion for AI alone to cure it."  That is certainly true, and Newsweek's headline went further than the actual article did.  There will be human doctors, but they are likely to be doing different things and doing those things differently.  And they will almost certainly be supported by AI.

We have to remember that, by definition, most doctors are average.  Maybe 10-15% are great, maybe 10-15% are poor, and the rest are in the middle.  The law of averages is not good that the doctor you end up with is one of the great ones.  Even if you manage to identify one, getting to see him/her is often a challenge.

On the other hand, all it takes is for someone to develop one AI that is better than the average human doctor.   Unlike human doctors, such an AI is not limited by geography or patient load. It doesn't sleep or take days off.  It could, in theory, see all the patients who want to consult it.  It may or may not be better than great doctors, but it certainly would be more available than them, and would be better than the average doctor you might otherwise end up with.

And that's why AIs will ultimately prevail for most of our needs.

We used to expect other humans to operate elevators and dial phone calls for us.  We used to have personal relationships with our grocers and butchers.  Technology and convenience eroded or eliminated these and other kinds of interactions, and that will happen in health care as well.

We often talk about the importance of patient-physician relationship, but when you rush to the ER or suddenly get surgery, competence outweighs familiarity.  If AIs can help get us better, we'll use them, and that opens up many opportunities for change.

Many protest that "health care is different," and won't follow the patterns of other industries.  They're often happy to try to just make the current system less bad.  That's too low a bar.

Let's re-engineer our health care system to take full advantage of the available and the coming technology, and to truly respect our needs, preferences -- and time.

Tuesday, May 23, 2017

Rise of the Drones

For those of us of a certain age, we expected to be living in a Jetsons-type world, complete with flying cars.  That hasn't happened, but it is starting to appear as though the skies may, indeed, soon be full of flying vehicles.  It's just that they may not have people in them.

Welcome to the brave new world of drones.
Many people may have viewed drones as a toy akin to radio-controlled airplanes (indeed, that's how they've been regulated).  We're beyond that now.  Last summer PwC asked "Are commercial drones ready for take-off?"  They thought so, estimating the total available market for drone-enabled services at $127b

Many companies have already been testing use of drones for various kinds of delivery.  Domino's, for example, has tested drones to deliver pizza, and Chiptole delivered burritos (in partnership with Alphabet).   

The company everyone is waiting for, though, has been Amazon.  They've already piloted Prime Air Service in England.  It offers a large but limited set of items, stored in a specially designed fulfillment center, while promising delivery within thirty minutes, as if you were ordering a pizza. 

Here's their video:

I don't know how crazy I'd be about having my new Fire TV delivered to the middle of a wet field, as happened to the customer in the video, but one takes their point.  If you already thought Amazon was fast, be prepared to think again.

Bloomberg reports that Amazon has now opened up a drone research center in France, aimed at developing their own flight control system.  It is one thing to program drones to avoid stationary landmarks like buildings or hills, and it's certainly easy to imagine using transponders to avoid other drones, but Amazon is thinking about "non-collaborative flying objects."

A.K.A., birds.                                                                

As Paul Meisner, Amazon's vice president for global innovation and communications, told Bloomberg, "Geese will never be collaborative so we have to sense and avoid those obstacles."  He admitted that there are many regulatory hurdles ahead, which may take years to fully resolve, but vowed, "We’re not going to launch this until we can demonstrate its safety."

This is not going to all be about getting your books, or your socks, or even your new HD television faster.  It is going to impact many industries -- including health care.

And that impact has already started to happen.

Zipline International, for example, is already delivering medical supplies by drone in Rwanda.  They deliver directly to isolated clinics despite any intervening "challenging terrain and gaps in infrastructure."  They plan to limit themselves to medical supplies, but not only in developing countries; they see rural areas in the U.S. as potential opportunities as well.  Last fall they raised $25 million in Series B funding.  

Drones are also being considered for medical supply delivery in Guyana, Haiti, and the Philippines.  

And drone delivery is already being tested in more urban areas.  The Verge reported that Swiss Post, its national postal service, is working with two hospitals in Lugano to ferry lab samples between them, which Swiss Post claims is the first commercial deployment of drones in an urban area.  Its press release claimed that "the regular use of drones between the two hospitals will become an everyday occurrence." 

Similarly, Johns Hopkins has been testing drone transport of blood supplies, concluding that it is "an effective, safe, and timely way to get blood products to remote accident or natural catastrophe sites, or other time-sensitive destinations."

Airbus is developing the A-180 drone specifically to deliver medical supplies, especially for emergencies.  Its cargo capsule is "capable of transporting everything from medicine and antivenin to supplemental blood and even organs."  A company called Otherlab is going a different direction.  Wired reports that their drone will deliver its package -- then decompose, making it ideal for deliveries to humanitarian crises (or to battle sites, since Darpa helped fund them).  

Lest we focus too narrowly on the concept of drones delivering medical supplies, argodesign has proposed a flying ambulance, which could be operated as a drone or by a pilot.  If you've ever seen ambulances stuck in traffic and felt sorry for the patients relying on them, such ambulances could be the solution -- arriving faster and to locations regular ambulances could not reach.  

Their concept:

But for real impact, let's go back to Amazon.  CNBC's Christina Farr broke the news last week that Amazon was considering getting into the pharmacy business.  They reportedly have been considering the move for several years, but now are starting to hire experts in the field, including a business lead..  They already sell various medical supplies and equipment.

Amazon knows prescription drugs is a complicated market, but one that experts and consumers agree needs significant change, due to high prices that are further obscured by various middlemen, hidden rebates, manufacturer coupons, and health plan discounts.  Stephen Buck, cofounder of GoodRx, told Ms. Farr, "I think Amazon would introduce a lot of transparency to what drugs really cost," estimating that it could be a $25b to $50b market opportunity for Amazon.

Even for Amazon, that's a lot of money.

Put rapid delivery -- especially with drones -- together with lower and more transparent prices, and it is no wonder that the stocks of CVS and Walgreens took a hit when the news broke about Amazon's new interest.

Dan Diamond writes in Forbes that Amazon's entry could be a game-changer.  He says: "for all of the major new players eying the health care market — with Apple pushing to collect health data through the Apple Watch, or Walmart beginning to deliver care at its stores — Amazon's innovative plan is arguably best-positioned to fill an existing gap."

Can anyone imagine Amazon would have much patience with PBMs like Express Scripts or Optum?

Kevin Schulman, a Duke professor of medicine, is intrigued by other possibilities Amazon could take advantage of, telling the Washington Post: “If Amazon would know that you have diabetes or hypertension they could do a lot with that data.  In principle, they could set up a marketplace where they behave differently, with different rules and different privacy practices.”

Health care has been all-too-much a story of waiting.  That's quickly changing, with telemedicine, WebMD, retail clinics, and -- soon -- 3D printing and health care robots.  We can add health care drones to the list, allowing 30-minutes-or-less kinds of promises that we haven't even begun to tease out yet.

Bring on the drones!

Tuesday, May 16, 2017

Picture That

Mala Anand, SAP's President and EVP- Analytics, recently wrote, "data is the fuel for the digital economy."  We generate more data than ever before, are finding new ways to derive value out of that data, and organizations in most industries are realizing that effectively applying that value is critical for success.  Health care included.  

Our problem, though, may not be in either generating or analyzing all that data, but in visualizing it.

Dataconomy used the example of the weather, for which we have huge datasets and highly sophisticated prediction models, all of which have to get boiled down into the slick graphics we've come to expect from our local weatherperson.

It is not enough to have the right data or the right data scientists; "you also need someone who has domain knowledge of your business and the ability to effectively communicate information back to end users."

Co.Design reported on new research from Autodesk that help illustrate (pun intended) how visualizing data is "a crucial part of analysis that can reveal surprising things about your data." The research takes 12 seemingly similar datasets that end up having very different graphical representations, providing insights that might otherwise have been missed.
Source: Audodesk
A recent Health Catalyst article, by Huesch and Mosher made the case for more data scientists in health care, noting that, among other things:

  • 30% of the world's stored data comes from the health care industry;
  • Of the approximately 6,000 data scientists in the U.S., only 180 work in health care;
  • health care could use 10 to 20 times more data scientists.
If you're having trouble with the math (perhaps you need a picture!), a sector that is 20% of our economy and has 30% of the data only employs 3% of its data scientists.  That sure seems like a problem.  

The authors outline of the barriers that have led to this shortfall, outline the buy-versus-build dilemma health care organizations face when it comes to beefing up their expertise, but believe that
Putting these pieces together will help the overall health care sector to achieve the same much-needed improvements in cost, outcomes, access, and experience that the data revolution has achieved in so many other industries.
It may not be easy.  Health care faces fierce competition for data scientists.  IBM recently profiled how fast the field is growing, with annual job openings increasing by 364,000 by 2020 -- 2,720,00 in total.  
Source: IBM
Unfortunately, the talent pool is nowhere near what it needs to be.  PwC, working with the Business-Higher Education Forum, urged that we address the skill gaps for data science and analytics: 69% of employers want workers with such skills, but only 23% of educators say their graduates have them.  They repeatedly cited data visualization as one of the core competencies needed.  

Having a bunch of quants produce reams of spreadsheets with statistically meaningful analyses of zillions of numbers is all well and good, but may not do much to improve anything we do, unless the decision-makers can understand them.  Health care already has plenty of statistics, many of which clinicians do not make full use of and which most consumers do not understand.  

Let's face it; most of us are not good with numbers.  Most of us don't think in numbers.  Most of us think in pictures.

Data visualization is a new form of visual communication, helping to provide insights into large datasets.  If you've seen an infographic, you've seen one form of data visualization.  The University of British Columbia provided this overview on the field:

A key statement: "it provides insight into complex datasets by communicating their key aspects in more intuitive and meaningful ways."

To date, most data visualization has been in 2D, seen on a screen or piece of paper, but we're already seeing efforts to portray data in virtual reality, such as by "creative science studio" Kineviz.  Holograms may be next.

Health care, despite its paucity of data scientists, is trying to embrace the data visualization.  For example, the American Academy of Family Physicians just issued their Vision for a Principled Redesign of Health Information Technology, describing their vision for how HIT can support improved care.  Data visualization was one of the first priorities listed, as they predict it will "...make it easy for the clinician to see patterns and make insight..."

Abhinav Shashank, cofounder of Innovaccer, sees data visualization as key to the future of health care:
Once physicians move away from long, incomprehensible data flows, and find an alternative that helps them instinctively read, isolate, and act upon the insights, only then can we be one step closer to a data-driven, value-based care. 
The University of Michigan Center for Health Communications Research, through funding by the Robert Wood Foundation, founded Visualizing Health,  A video explains their purpose:

Their interest is not just academic; they want to help people do their own data visualization.  They provide a toolkit for consumers and organizations to better display data, including The Wizard and a gallery of visual approaches to data.

Health care is desperately trying to reshape itself from a hands-on, more-art-than-science, physician-centered enterprise to a data-driven, value-based, patient-centered science.  We're not there yet.  Big data is expected to play a crucial role in this transformation, but, as Sutter Health's CHIO Sameer Badlani recently said, "Big data has moved on from infancy.  It's in the terrible twos right now.  We're still trying to figure out what to do with it."

A large part of that has to be how to explain all that data to its various users -- practitioners, executives, and consumers.  Data visualization will be key.  Health care may or may not need more doctors, but it certainly needs more innovative business models, better technology designers, and more data and computer scientists.

And data visualization experts.

Google has invited designers and artists from around the world to tell better stories about Google data through data visualization.  What health care organization is ready to do the same with their data?

Wednesday, May 10, 2017

Ask More, Listen Better

A new study in JAMA suggests that nearly one-in-three drugs approved by the FDA between 2001 and 2010 had post-market safety issues, which caused safety communications to physicians and consumers, "black-box" warnings on labels, and drug withdrawals.

It is not clear how many patients may have died or otherwise harmed by these issues.

Some complain the FDA takes too long to approve new drugs, but Kaiser Health News pointed out that the same Yale researchers had previously found that the FDA actually approves faster than European counterparts, and the study found that clinical trials typically involve less than 1,000 patients, and usually for less than six months.

Lead author Joseph Ross, M.D., noted: "No drug is completely safe, and during premarket evaluation, we are not going to pick up all the safety signals," and urged "that we have a strong system in place to continually evaluate drugs and to communicate new safety concerns quickly and effectively."

Dr. Eric Topol, who was not involved in the research, told the Washington Post that he was troubled but not surprised by the findings, and similarly suggested: "Why not have a standard where we put every new drug under watch, and see if we could catch a problem before the drug is widely advertised?"

Why not indeed?

In fact, why just new drugs, and why only drugs?

Take medical devices.  The FDA has a formal process for medical device reporting, which they say results in "several hundred thousand medical device reports of suspected device-associated deaths, serious injuries and malfunctions."

That sounds ominous.

Still, though, these reports require that the manufacturer report instances when their device "may have caused or contributed to a death or serious injury," which requires that someone -- a facility, a doctor, patient, etc -- tell them.  Of course, deciding what "caused or contributed" to a death or serious injury is probably as unclear as what entails a "serious injury."

Last year the FDA cited 12 hospitals for failing to report such issues, while also warning that "we believe that these hospitals are not unique in that there is limited to no reporting to FDA or to the manufacturers at some hospitals."  Reporting by the actual patients and their families is voluntary, and most of us probably would not think of doing so in most circumstances.

Even worse, as Elizabeth Rosenthal profiled in An American Sickness, manufacturers can sometimes avoid FDA approval entirely, as happened with, for example, hip implants and surgical mesh.   Such problems may never be reported, or only after damaging enough patients that someone finally realizes there is a problem.

One has to believe more doesn't get reported than does.

This lack of ongoing oversight is a pervasive problem in our health care system.  Ever read those warning labels on drugs, which detail all the potential side effects?  Do they make you feel better, or worse?  Do they help you understand how likely you are to have them?

Chances are, none of the clinical trials had patients with your specific set of health issues or with your exact combination of other medications, so it's anyone's guess how you might respond to a drug, new or old.

More to the point, if you do suffer any side effects, what are you supposed to do?  You may just accept them.  Or perhaps tell your doctor, who maybe switches drugs, or maybe not.  In any event, probably no one is tracking or reporting the incidence of most such side effects.

This is not just a problem with drugs and devices.  Maybe you had a surgery.  If you ask your doctor before the surgery how back to "normal" you can expect to get, when, chances are he/she can't tell you -- and if he/she does, you'd be well-advised to question the source of those numbers.  There may be statistics on, say, how many patients get re-hospitalized and/or get an infection, but as to when you'll be walking without a limp, it's pretty subjective.

We track loads of "quality" metrics and conduct numerous patient satisfaction surveys, but not many people believe we're actually measuring quality, much less how a specific patient is doing today.  Nor do we have any firm idea what that patient should expect to feel like tomorrow.

Our mechanisms for tracking how patients are doing after we do something to them are minimal at best.  Physicians tend to rely on patients calling with any problems and on follow-up visits, but both leave lots of cracks to fall through -- and even they do not usually end up being recorded in any useful way.

This kind of "squeaky wheel" reporting is antiquated.  It comes from an age when it was hard to effectively track how people were dealing with their heath issues, and impossible to make sense of the mass of data that would be generated even if it was collected.

None of that is true now.

We have wearables.  We have online surveys.  We have medical device registries.  We have automated calls and texts.  We have patient review sites like Yelp (which may do better than formal surveys in tracking patient issues).  We'll soon have Internet of Things options that we can barely even imagine, from devices we won't even realize are there which are "listening" to us 24/7.

We should be able to track almost anything we wanted to about how a person is feeling, in real-time or near real-time.

No human could track all the data that will be generated, much less already-to-busy physicians.  That's OK; we have AIs that are becoming more and more able to sift through all this data and drawing meaning from it.  AIs could quantify the likelihood that you would get a side effect from a new drug, determine that you are suffering from a side effect from that drug, tell you when you should be able to walk how far after your knee implant.   They'll be able to alert your physician when something appears off.

Our current approach -- relying on someone to complain about problems -- is a way to uncover problems, but it is not nearly as effective as actively collecting and analyzing the data that would reveal the problems.  The good news is that we now can.  The bad new is that we are not.

We need to ask more, and listen better.

Tuesday, May 2, 2017

How About Some Meatballs?

You've probably heard of IKEA, the international retail giant that is known for its modernist design aesthetic and its clever some-assembly-required furniture.  You may have wandered around one of it cavernous stores or received their catalogs.  You may even have guessed that IKEA is the world's largest furniture retailer, with some 400 stores in almost 50 countries and annual sales of over $36b.

What you may not realize that IKEA's restaurant business has annual revenues of $1.8b.

Fast Company reported that IKEA has started taking its food division -- which includes cafe/restaurants and Food Markets -- more seriously.  Michael La Cour, the division's managing, told them:  
This might sound odd, but it’s almost something we didn’t notice.  But when I started putting the numbers into context of other food companies, suddenly I could see, well, it really is not that small.
At the time, in 2013, those revenues were $1.5b, but by 2016 were up to $1.8b.  OK, that's not McDonald's territory, but well-known chains like Five Guys, Steak-and-Shake, or P.F. Changs would salivate at those numbers, or at the 8% annual growth.  IKEA serves some 650 million diners a year, and are known for their family-friendly prices.

And, of course, for their meatballs.

Business Insider calls them "one of the most underrated restaurant chains in America."  LA food critic Merrill Shindler noted that the Burbank IKEA's restaurant holds 800 diners at a time, "But show up on a Saturday at lunchtime and perhaps twice and even three times that many are lined up, cheerful and flushed with the joy of consumption, and hungry for Swedish meatballs."

Oh, two other facts that should be of interest.  IKEA estimates that 30% of their restaurant customers are coming there just to eat, not to shop,  The cafes may have started as a way to help keep people shopping longer, but they've become a dining destination for some customers.

Which leads to the second interesting point: they've tested standalone pop-up restaurants in several cities, including London, Paris, Oslo and Toronto.

Curbed reports that,IKEA demurred on expansion plans for its food division, saying no decisions on standalone restaurants have yet been made, but one would have to guess that it is much less expensive to open up an IKEA restaurant than a new IKEA store, with pop-up restaurants especially cheap to test.  

And then I think about hospital food.

That's not entirely fair, of course.  Many hospitals have made efforts to upgrade at least the appearance of their cafeteria, to make the food selections healthier and more appealing, and to give patients more control about what their choices are.  Still, though, as Dr. Joel Kahn wrote in the Huffington Post last year, "There is wide agreement that the quality and choices of food at most hospitals is abysmal."

The Physicians Committee for Responsible Medicine scored the food menus of multiple hospitals, finding significant variation in how healthy their offerings were.  In the previous year's report, they detailed how many hospitals contract with fast food firms under terms that they say helps promote unhealthful foods.

It is not a pretty picture.

Two things I'm pretty sure of: unlike IKEA's cafes, no one is choosing to dine in a hospital cafeteria, and no hospitals are thinking about opening their own free-standing restaurants, pop-up or not.

To be fair, institutional food is hard.  Not many people like school lunches or find stadium food a good value, and let's not even talk about prison food.  In hospitals there is the additional problem of trying to balance the needs of sick patients versus the tastes of their presumably healthier visitors.  The end result is often food that no one likes and that no one would stay longer for, much less come for.

You have to wonder how IKEA would address the problem.  Or, as they might see it, the opportunity.

Many retailers offer some food options.  Still, though, no one is praising Walmart or Costco food, and most of the cafes that large department stores used to offer have gone by the wayside.  IKEA knew it had hungry customers and wanted to help them with that, managing to do so with good quality and affordable prices.

It wasn't their business but they made it their business, and did it well enough for it to become a business on its own.

There are two distinct lessons that health care organizations should learn from this:

  1. If you're going to do something, do it well.  Hospitals have to serve food, and their "customers" don't have much choice about eating it, but that's no reason to do it badly.  If a furniture retailer can make a dining experience delightful, why can't a hospital?  
  2. Be open to opportunities: IKEA wasn't really in the food business, much less the restaurant business.  But once they realized its potential, they weren't afraid to embrace it.  No "but we're in the furniture business!" objections.  They were good at something, customers were seeking them out, so they didn't let their past dictate their future.
This is not a suggestion for hospitals to get in the restaurant business (although it is a plea for them to at least do their food services better!).  It is, rather, a suggestion that health care organizations -- not just hospitals -- do a better job of identifying and taking fuller advantage of their assets and opportunities.  

Hospitals have been on an consolidation/acquisition spree, both horizontally and vertically.  Much of that, though, seems to have been more about locking up their existing markets than opening up new ones.  Certainly we have hospital systems like UPMC or the Cleveland Clinic expanding internationally, many health systems opting to get in the health insurance business, lots of hospital-run health clubs, and much investment in new technologies.  

Still, though, how many are doing something that really makes you think "well, that's interesting!"?  Like IKEA's expanding into the restaurant business.  

If meatballs prove to be IKEA's breakthrough asset into a new business, which of their assets are health care organizations overlooking?