Consumer-focused healthcare can save lives by focusing on changing behavior

Everything we do in the $3 trillion healthcare market today only affects 10% of outcomes to premature death.

You read that right. All of that, for just 10% of outcomes:

That 10% exists for a reason. Genetic predisposition is hard to change. So, unfortunately, are social circumstances and environmental behavior. But that 40% of behavioral patterns — why can’t we tackle that? This is what real prevention would look like: nothing comes even close to mattering as much towards whether you will die prematurely as your behavior does.

We can do better than simply focusing on that small 10% slice of the pie; in fact, we’re looking in the wrong place. Doctors, entrepreneurs and founders need to be thinking (and treating with) lifestyle as medicine. Because behavioral change is the best and most powerful way to impact that whopping 40% slice.

Too often we think of this as the “just eat right and exercise” problem. As we know very well, that platitude will not solve our healthcare problem. The true problem is the difficulty of modifying behavior. We know this, because the platitude doesn’t work. We like to eat what we want, to exercise or not exercise if we choose. In short, humans like our patterns. They’re hard to change.

Tech, on the other hand, modifies behavior very well. Just look at the phone you’re probably reading this on, which has foundationally changed the way we communicate — along with huge other swaths of human behavior, in both positive and negative ways — from the ability to call a ride service in practically any city at any time to tracking your health to screen addiction. We know technology modifies behavior; we live this every day. So the question is, how can we target this superpower ability of tech to have 4x the ability to impact that the $3 trillion healthcare budget does?

How does it work?

Let’s think about why technology actually does work for modifying behavior. For one, it’s always there, thanks to the leap in mobile tech, whether that be phones or fitness trackers. Second, technology’s ability to do constant A/B testing essentially enables RCTs, or Randomized Clinical Trials, every moment that technology is present and being used. These RCTs are invaluable laboratories for learning about what is effective therapeutic behavior modification, or improving efficacy — and it’s not toxic. Most medical products are released and then rarely get updated (think about how old the stethoscope is!). Rolling out new versions of products has been difficult and expensive. But that no longer has to be true. The same kind of A/B testing that Amazon does, for example, to optimize ecommerce — everything from the look of the website to the flow of the experience to the nature of the shipping that you get — can be now applied to behavior modification for health. Comparing the immediate efficacy of two algorithms for lifestyle behavior modification on two different populations can happen not just over years or months — as a RCT would have to be — but over weeks and even days, improving our responses and lifestyles that much faster.

Second, applying Machine Learning to vast amounts of new data is identifying all kinds of nuances of human behavior that we aren’t nearly as good, as humans, at noticing. For example, correlating patterns with data like where you shop, when you eat lunch, what activities do you do, what shows you watch, what your exercise routine has been, how much you sleep, even perhaps whether you remember to charge your phone. Identifying the clues in our behavior that eventually add up to significant lifestyle risk is the first step towards changing and improving that behavior. Like it or not, we live our lifestyles now through our phones — ML allows us to learn from it.

And last, technology allows us to scale existing therapies in new orders of magnitude.  Programs which have proven extremely effective at behavior modification through personal interaction — such as Diabetes Prevention Program for Type 2 Diabetes — have been by definition hard to scale; computation can extend their reach into the billions. Or take for another example depression, a complex disease where the molecules involved are poorly understood: drug therapies have been challenging, but therapy, specifically CBT, has a very strong track record, and computational CBT — ie, CBT scaled with technology — the strongest.

Even conditions as mysterious and difficult as cognitive decline can be treated much more effectively with technology. This is another fascinating example where the biology is so complex at the molecular level that breakthroughs have been far and few between. On the other hand, cognitive is painfully clear at the behavior level. And it is also very clear that behavioral treatment in the form of cognitive stimulation helps significantly. In this study, for example, the auditory memory and attention capability of patients who received cognitive stimulation training 1 hour per day, 5 days per week, for 8 weeks improvement was significantly greater than those who did not.

These are big challenges to meet. Behavior is the result of thousands of small decisions at every moment of every day: do I sit or do I stand? Do I drink this beer? Even, do I take regular deep breaths? One of the biggest challenges to face is how we ‘read’ this behavior and turn it into reliable data. There’s also the issue of small sample sizes: in order to narrow down to a meaningful experiment, you need, at the moment, to have very clear definitions of behavior, which often means small sample sizes of people who always do X in Y conditions. The science of behavior and decision making itself is complex, debatable, and often evolving. And there’s the company building practicalities: to build a company in this space, you need to find people who understand clinical science, data science, experimentation approaches, behavioral science *and* product and UI.

But that’s exactly the opportunity. These things are coming; we understanding more about behavior every day, as devices enter our daily lives and health data becomes more and more fine-grained. New conceptions of roles that blend behavioral science and product design are clearly emerging. All of these means are not exclusive and can be combined into powerful ways of modifying behavior for health. Those that can connect all these dots have the ability to build companies that can take a giant bite out of that 40% — and have tremendous impact on mortality for huge swaths of the population.

There’s an old joke that plumbers have saved more lives than doctors, because improving sewers and sanitation (and eradicating the disease that went along with that) was so impactful on longevity for humans. By cleaning up the modern day ‘sewers’ of our lifestyles — not through magical drugs, complex procedures, or platitudes about prevention — but through a real infrastructure of technology that is being built right now — technology will bring an analogous impact.