Data is only useful when it leads to better decisions and behavior. While many of us collect and possess troves of health-related data about ourselves, I think we’re in the very early stages of extracting value from that data.
I spend most of my time these days around pro athletes and former athletes, a health and performance conscious bunch. The majority use some kind of personal tracking device—Whoop bands, Oura rings, Apple Watches, FitBit, etc. Many of us also log our workouts either by hand, or with an app like Strava. While the wearable and training data are fun to look at, they don’t often change our behavior, or better yet, facilitate some action on our behalf. We are collectively succeeding in data acquisition, but failing to use the data to make ourselves healthier and fitter, the ostensible goal in collecting all of this data.
In many parts of the digital world, past data is used to make a decision about the present, whether that’s deciding which ad has the highest probability of getting clicked, or how to translate a phrase from English to Spanish. This is the allure of self-driving car technology: use past data to train a system that makes better decisions than humans, in a realm where the downside of a bad decision can be fatal.
Current State
Unfortunately, we haven’t reached the point where we can use our health data in the same way. Even if we’ve collected all the personal data we need in order to know what we should eat, how we should exercise, or when we need more recovery, the data is often in disparate systems, with no central layer aggregating, learning, and recommending the best action on our behalf, greatly reducing the value of any data we’re collecting.
With existing technology and standard lab tests, we can track our sleep, the physiological load and intensity of our workouts, how recovered we are, how different foods impact our glucose levels, all of our hormone levels, etc. On their own, all of these datasets are interesting and potentially helpful, but the biggest value comes through integrating them together.
While apps like Gyroscope and Apple Health connect some of these disparate datasets, it’s still difficult to answer questions like how does what I eat affect various hormones and energy levels over time? What’s the optimal food and timing to eat before an afternoon workout? How should I change my workout after a poor night of sleep? The most dedicated trackers can get these answers by spending a lot of time wading through and matching raw data files, but most will never go to these lengths. Eventually, I think we’ll see a centralized layer of our personal health data that provides very specific, personalized insights that answer these important questions.
Recommended Actions
Once we have a common layer where our data reside, the next logical step is a system that recommends and automates decisions based on our personal data. In the current wearable and testing ecosystem, our data live in separate vacuums, behind the walls of the testing or tracking company, which leaves the individual responsible for connecting the dots, and making behavioral changes based on the data.
I think there can be a better way; instead of a system that decides which ad to serve, we’ll be interacting with a system that takes as inputs our sleep data, most recent blood tests, workout data, glucose levels (and whatever other data we have), and produces an ongoing series of recommendations for us:
- an auto-generated Instacart grocery order each week consisting of only foods we know we tolerate well and supplements that our blood tests suggest we could benefit from. With a click, we confirm the order.
- a daily workout that responds to our most recent workouts, diet, and recovery metrics
- on nights we need more sleep, the system triggers our phone and computer to require 2-factor authentication to access past a certain hour, and our lights automatically dim earlier in the evening
The list of potential use cases goes on, but the point is that the insights that already live in the data we have today actually turn into actions that improve our health, mood, and productivity. Humans have a long history of acting with amazing irrationality regarding their health, and technology like this can help our plight.
Examples of technology using personal health data to recommend an action already exist. After a suspected fall or elevated heart rate for an extended time, Apple Watch will prompt you to make an emergency call, which has likely already saved many lives. How do we get technology to take similar actions for us in more benign, non-life-threatening scenarios?
Where to next?
The progress of companies that are built on recommendation engines is staggering–Google, Netflix, Spotify, and Amazon are all world class in predicting what you want to see, hear, or buy next. While Google has an online fingerprint of all of us, there’s not yet a great physical fingerprint–just disconnected smudges that give partial information about us. Once these various pieces coalesce, technology’s ability to provide better lifestyle guidance will reach new heights.
If our goal in tracking biological and physiological data is to feel better, be more energized to do the things we love, perform better, and have more mental clarity, then we have to understand how these various biometrics connect together, and how the lifestyle choices we make impact the quality of our lives.