Imagine getting a text message on your smartphone that says: “Hi! You’re approximately thirty days from a heart attack. To find out what you can do to stop it, please click on this link.” Would you click the link? More fundamentally, would you want to know about a major health problem before it happens?
From the times of Damocrates, the Western way of medicine has been reactive. We go about our daily lives until a health problem forces us to walk, run, or get carried to a doctor. The harried physician tries to figure out the likely problem based on a quickly gathered set of symptoms. In many cases, the doctor is given little to no understanding of our medical history or lifestyle — no sense for the complex interplay of our most recent ailment with last year’s broken leg, last week’s work-related stress, or last night’s third beer.
One of the great ironies of our digital age is that our physician knows less about us than our social networks, our credit card companies, or our dating apps. The human body is a highly complex piece of machinery and its health — our health — is affected by hundreds of daily micro-decisions from the sleep we interrupt, the morning snack we eat, the news we watch, the coffee we drink, the exercise we forego, all the way to how we react to an angry customer at work. Expecting a doctor to get an accurate picture of what’s wrong and how to fix it is like asking a mechanic to fix a broken car based on a one-time Polaroid of the engine. In fact, it’s nothing short of miraculous what our doctors can do given that they know so little of the realities of our daily lives — imagine what they could do with a full picture.
The idea that health is far more than a collection of vital signs, lab results, and diagnostic images is not new. However, what is unique about the here-and-now is our technical ability to record and make sense of the constant data streams that our bodies produce.
Recording the data was the first Rubicon. The rise of wireless communications and the miniaturization of sensors in the early 2000s gave rise to the promising field of telehealth. Bluetooth-enabled blood pressure monitors and weight scales coupled with videoconferencing meant that the doctor could now come to your home for a virtual visit. Perhaps more importantly, it meant the doctor could see how your blood pressure and weight fluctuated over time — that is, so long as you remembered to step on that scale or take that blood pressure measurement.
We took another huge step forward with the rise of connected devices and consumer wearables. Now, you didn’t need to remember to step on that weight scale after all. Just pop on your Apple Watch or Fitbit and go about your daily life — the passive sensors inside can automatically track your activity, your inactivity, and, eventually, more and more of your vital signs.
Of course, the challenge with recording so much activity data becomes one of processing. If we present our doctor with tens of thousands of data points of walking, running, exercising, sleeping, heart beating, and so forth, it would likely require a good deal more data analysis than possible in the twenty minutes allotted for our appointment.
Even if our doctor had the time to analyze all of the peaks and valleys in our data, the Thanksgiving lulls, the post-New Year resolution activity spikes, and everything in between — there would still be the problem of ensuring the data could be superimposed over the appropriate baseline. A baseline activity level for me looks rather different than the baseline for an Olympic athlete. What is the right comparison cohort, then? A busy doctor can maybe see a thousand patients a year. That’s a limited panel from which to make statistical inferences.
Fortunately, the processing problem is being addressed at a higher level. Analytics companies have long worked with health insurers to conduct data analysis for populations of millions. Traditionally, much of that work centered on claims data — the health insurer’s bread-and-butter — and sought to identify how a patient’s diagnoses, procedures, and prescriptions are correlated with the risk of future health problems. This sort of analysis was useful in terms of big-picture population averages, but the real breakthrough happened when new challengers in the analytics space started looking at less traditional datasets. “Wait a second,” they said, “how is it that a patient’s zip code can sometimes tell me more about their likelihood of that patient ending up in the hospital than their medical record? And how good is knowing a patient’s prescriptions when that patient lacks money for transportation and can’t get to the pharmacy to have them filled?”
It turned out that the analytics companies were picking up on something that care managers and caseworkers had been seeing all along — patients’ health was based as much on their clinical data as it was on their socioeconomic and environmental data, the so-called “social determinants of health”. The interesting part was that connected devices could fill in a lot of the blanks, providing both clinical and social data. By putting it all together with traditional claims and assessment data, we are starting to assemble the first holistic picture of the patient and knocking on the door of whole-person care.
We now have the machinery in place to collect richer datasets than ever. The consumer apps tracking activity and sleep are starting to be integrated with clinical information and caseworker notes about the social determinants of health. At the same time, the advent of machine learning means that we can discover new correlations previously unknown to medical science and use them to inform treatment protocols that go well beyond taking two pills in the morning and calling the office if you don’t feel better.
Indeed, we have arrived at the sunrise of digital therapeutics — the inevitable idea that it will be software telling you when and if to take that pill, or forego that extra cup of coffee, or get that extra hour of sleep. Digital therapeutics augment the doctor and the traditional prescription: they can follow you home, speak to you at work, and monitor you at the gym. You are already discussing your choice in music with Amazon’s Alexa and querying Google Assistant about the weather — why not have a chat about your health?
As we collect more and more data, and our machine learners get smarter and smarter, our digital therapeutics and analytics will inevitably get better at predicting what health events are likely to happen and when for each patient, and what can be done to prevent or minimize their impact. Of course, data privacy concerns remain paramount. Fortunately, we have a robust regulatory framework in the United States for the protection of personal health information. So long as the regulators ensure that our data is used for prevention and treatment rather than raising our insurance rates or lowering our salaries, I’m happy to provide all the data points I can. When given the choice between a text warning me of a heart attack and the attack itself, I’ll take the text, and the chance to improve my health outcome.
Technology is revolutionizing the Western way of medicine and taking us from reacting to preventing. There is still a long way to go, but the proofs-of-concept and the pilots that are already out there are nothing short of breath-taking. It’s definitely an exciting time to be alive and thrilling to be part of the cutting-edge developments that will keep us that way.
Jack Plotkin is the CEO of Cardinal Solutions, a boutique advisory and investment firm based in New York City. He has more than two decades of experience at the crossroads of business and technology and has advised more than a hundred Fortune 500 firms across all major industries, including healthcare and life sciences.
Originally published at http://professionaltales.com on September 17, 2019.