Why Big Data has Yet to Revolutionize Healthcare

Article originally written in 2016 and we’re still waiting.

Everyone is claiming that we stand at the precipice of one of the greatest shifts the health world has ever seen. Big data is going to change everything: How we diagnose, how we treat, how we monitor, how we live, and how we die. Yet while optimism abounds within the tech community, many who have been long standing in healthcare feel as though they’ve been sold a false bill of goods.

And they’re right. Big data has not become the savior to the health industry that we were all expecting, nor will it any time soon. Though the two sides to this debate are strongly rooted in their positions, they are both fundamentally wrong in their conclusions. Most technocrats believe that technology will solve all of our problems, and it is simply regulation and blind conservatism standing in the way. Meanwhile, many medical traditionalists feel that technology and big data will never replace human intuition and the institution of doctors.

The sad reality is that the answer lies somewhere in the middle. To overcome the hurdle preventing us from evolving the health system, we require the cooperation of both sides. The fact is, we have absolutely no idea what we’re looking at. That’s the true barrier recognizing big data as the holy grail it could be. We lack a baseline understanding of consistent, longitudinal health data that allows us to interpret and understand a medical condition.

Think of it this way: The Hippocratic Oath is over 2,000 years old, and modern medicine is roughly 200 years old. It took us hundreds—if not thousands—of years to evolve and develop our understanding of the human body through empirical methods and the tools at a doctor’s disposal.

In comparison, the big data movement has only truly emerged within the last decade. I will gladly concede that data has been used to study the human body for ages; however, never in our existence have we had access to such reliable, consistent, and accurate data to analyze. When it comes to our understanding of medical big data, we are in the technological equivalent of the Bronze Age. The Human Genome Project dates back to the mid ‘80s, and we have yet to see great progress in genetic therapies or DNA tampering.

Much like our current view of archaic health practices of yore, our current efforts will be looked back upon as foolhardy mistakes. Products like the Nike+ FuelBand or Fitbit will one day join the ranks of practices such as leeching or trepanation. Increased professional medical efforts will be seen as missing the point, because we’re asking the wrong questions of our data.

In fact, we don’t even know the right questions to ask until we establish a baseline for what “normal” human beings look like under the big data-microscope. We can collect heart rate, temperature, weight, body chemistry, and EEG data until we’re blue in the face, but without a frame of reference, this data becomes little more than a novel distraction from meaningful medical analysis. Until we have a better understanding of what can be considered normal—across billions of people, mind you—and what is a concerning deviation from that norm, our data doesn’t do us a lot of good.

For example, if we monitor every single heartbeat, we are bound to see abnormal rhythms emerge in all of us at least a couple of times a day. This does not mean that everyone has cardiac dysrhythmia and should get pacemakers. Similarly, at some point throughout the day, most of us likely have blood pressure which rises temporarily into dangerous levels due to stress, activity, or nutrition. However, this does not mean that we’re all at risk of a stroke.

Moreover, we need to be able to understand the difference between the causation, correlation, and irrelevance of data when studying certain conditions. Cold body temperatures may have direct causality towards hypothermia. Warmer body temperatures may have a correlation with the flu. Rapidly fluctuating body temperatures, however, may have nothing to do with either; we might just be exercising or entering cold spaces. We should not let data guide us blindly down questionable rabbit holes. Instead we should ask intelligent questions of the data to determine possibly health implications.

And lastly, we must be conscious of placebo effects and the uncertainty principle in all of this. If data is in fact enabling paranoid, obsessive individuals to turn into statistically-armed hypochondriacs, we have potentially created more problems than we have solved. While the data could eventually be meaningful and insightful, we need to have the right individuals interpreting the data in the right way—not simply reacting to every deviation from the norm.

In many ways, our ability to gather biological data has surpassed our understanding of medicine. We’ve never before been afforded this kind of all-encompassing access to the human body. Like newborns entering the world, we’re still trying to figure it all out. We’re over stimulated by the presence and quantity of data around us, and much like addicts, we lack the sophistication and self-control to know how to use this resource rationally to our advantage. We also risk of bringing ourselves serious harm by losing sight of what is most important in what we’re trying to accomplish.

While our understanding of big data will eventually catch up to our ability to collect it, it’s imperative that we never lose sight of treating the individual—not the numbers. We should never fetishize data or hold it in higher regard than the individual themselves, how they feel, and who they are.

Big data, like the stethoscope, the electrocardiogram, and the thermometer, is just a tool. While it could be a tool to revolutionize the way that we diagnose and treat medical conditions, we must never put the tool ahead of the task.