The Black Box of You: Why the Quantified Self is so Frustrating Today

The Black Box of You: Why the Quantified Self is so Frustrating TodayImagine a black box.

You can feed all sorts of information and data into it all the live long day. But the amount of data you can get out of it is limited. It just stares back at you with its blank, neutral sides. It can tell you things like where it was manufactured, or what color it is, or how many times it’s been moved today. But it can’t tell you what’s inside it, and whether what’s inside needs a little attention.

That is the human body today — you can put things into it, but the amount of information you or I can glean from it is limited.

It’s a frustrating situation, made even more so by the quantified self and personal genetics movements.

How could a personal data movement regarding your health or your body actually make things more frustrating?

In short, because the data we can get from such devices and technology is so limited, especially when compared to the vast richness of data that resides in our bodies.

Personal Health Trackers

Let’s look at some of the personal data trackers available today. One of the more popular ones is called Fitbit. It measures the amount of activity you engage in (e.g., walking or running), and your sleep patterns. By measuring activity, it can also indirectly measure calories burned. But because it is an indirect measure (e.g., it isn’t measuring your heart rate or oxygen intake), the calorie measurement is simply a reasonably accurate estimate. But an estimate nonetheless.

Knowing how many calories you burn is great when you’re walking, jogging, or engaging in some sort of exercise. But we also burn calories when we’re idle, and your calorie burn rate can differ a great deal between watching television, or working at the computer. And if you’re having difficulty with sleep issues, knowing your sleep patterns can also be extremely helpful.

Now, I don’t mean to pick on Fitbit — all of the personal health trackers available today suffer from these same intrinsic flaws.1

Personal Genetics

Personal genetics is another category of the quantified self movement that seeks to better understand the machine that is our human body. And the amount of information that a company like 23andme can provide an individual is pretty interesting. The current kit, for only $99, will provide you with over “200 health and traits reports.” They were originally focused on health, but have moved into ancestry in a big way, seeing the popularity of exploring one’s roots.

The ancestry portion tells you your: ancestry composition, genetic relatives, maternal & paternal (if male) lineages, and Neanderthal percentage.2

On the health side of things, 23andme will tell you your carrier status for 50 diseases and disease risk for 120 conditions. They can also tell you some traits about yourself (such as your hair and eye color — something a mirror does equally as well), and how well you’ll respond to some types of medications.

However, what 23andme doesn’t make very clear to you is that a lot of these traits and markers and such are based upon only one or two studies. As anyone who knows my writing over the years, I don’t draw any conclusions as a researcher based upon a single study. A single study tells you, “Hey, this is potentially interesting.” A followup study — from a different set of researchers preferably — confirms the initial research and lets you know that initial research may be robust.3

I didn’t go through and click on every trait and condition. But on a sample of 20 I did click on, the vast majority were marked as “preliminary research report” on one or two genetic markers. 23andme tells you that this kind of report:

Preliminary Research includes results of studies that still need to be confirmed by the scientific community. It also includes topics where there may be contradictory evidence. The results of these studies are not conclusive.

In other words, a lot of what 23andme is sharing with you may not actually be confirmed in the research. They’re sharing data with you that the scientific community hasn’t yet agreed on or come to any conclusions about. And I can’t say this was for the majority of their health conditions and traits, but it seems like it was for an awful lot of the random ones I checked.

The Black Box Syndrome

A few years ago, I was diagnosed with borderline cholesterol levels. It was a warning sign from my body — shape up or face potential future health problems.

Which was great information to have. But the only way to get a reliable cholesterol reading today is from a doctor-prescribed blood test. Which is generally a pain in the ass to get — and here’s the key — to get regularly.

Today, we’re in a situation where the black box can tell us where it came from (and what color it is), but not how it’s actually doing. It can share with us proxies for the real information we want to get. But those proxies are sad substitutes for the actual data we want from our bodies.

I already know how many calories (on average) I burn a day and I don’t have any sleep problems, so something like the Fitbit is of limited value and interest to me.

Because of the outstanding research issues, I think personal genetics is something also of limited value. It can maybe provide you with a possible heads-up or warning about potential future issues in your life. Or it could give you information that only makes you more anxious. Even if you know you have a 1 in 10 chance of getting XYZ disease, in most cases, there’s not a whole lot you can do to change that odds-ratio.4

I don’t want to know any of this. I want real information taken directly from my body. I want my cholesterol levels, in real time, like glucose monitors. I want to know the various levels and activity of my neurochemistry, to track patterns of energy, thought and attention. I don’t want a heart rate monitor, I want a heart health monitor — something that tells me when my heart is doing great, or when it’s got some issues that need further attention.

Heck, I want all of my vital organs monitored — because they are what keep me alive! I want a “cancer detection monitor” that can look for the very earliest signs of cancer and let me know it’s time to go see a doc. I want real-time monitoring of my own pacemaker (if I had one). This is my body and the data it provides should be mine.

Like a “check engine” light in a car warns you to take the care to a mechanic soon, I want a “check engine” light for my body, warning me when I should get in to see a doctor. Like most drivers, I don’t want reams of indecipherable data that I would need a PhD in mechanics to understand. I just want a simple dashboard that gives me live, real data on my body’s internal functioning. Not proxies and estimates of data.

Today, the quantified self movement is a far cry from something so simple and elegant. While knowing how many steps I’ve take today is great (a free pedometer from McDonalds did the same thing nearly a decade ago), knowing the progress I’m making on my cholesterol levels is far more useful and actionable data.5

When cars were first invented, they didn’t have a fuel gauge. Drivers needed to estimate on their own how much gas they had left by doing some easy calculations on a piece of paper. Now car dashboards not only tell you how much fuel they have left, they’ll tell you pretty accurately how much farther you can go with that fuel. They can alert you when it’s time to change the oil, or that your brakes need checking.

Today, we’re still doing the simple calculations on paper about our bodies, making guesses and estimates based upon proxy data (because the real data is too hard to get at regularly, and often without a doctor’s visit). I envision a future where we can replace such rudimentary measurement with a refined, simple health dashboard — that gets its data automatically from our bodies. Wouldn’t that be great?


  1. And exploring these gadgets can get expensive… For example, if you drop a Fitbit on a hard surface, there’s a good chance you’re going to break it (and be out $100). []
  2. Neanderthal percentage is a perfect example of information that I think no one, ever, asked for. []
  3. Ideally, we wait for even more studies, but hey, I’ll take two studies and call it the beginnings of a “robust” finding. Popular writers like Gladwell sometimes seem to do this as well, so it is unfortunately becoming more commonplace. []
  4. Yes, there are some notable exceptions, like Type 2 diabetes. []
  5. If you don’t think so now, just wait until you get older… []

Posted in: general | PM Tech | reforming hc | shared decision making





9 Responses to “The Black Box of You: Why the Quantified Self is so Frustrating Today”

  1. Hi John,

    I completely understand your frustration with the quantified self movement, and I think you have made valid points about the limitations of personal health trackers and personal genomics. But I think there is another category of ‘quantified self’ activity that can be of value, especially for chronic e-patients: symptom tracking.

    Tracking our symptoms and the lifestyle elements that may affect those symptoms (food, exercise, etc) can help patients better understand what behavior changes they can make to improve their quality of life. We know that the simple act of tracking creates a state of mindfulness that can bring about behavior change; it can also help patients solve medical mysteries that remain chronically unsolved by traditional medicine.

    Case in point: I have struggled for years with gut issues. and then last year I started experiencing severe burning in my hands and feet. It would come and go, seemingly according to what I had last eaten but I wasn’t sure. I avoided tracking my food and symptoms, thinking it would be too much of a chore. Luckily I was bombarded by the founders of Mymee, a tool created specifically to help patients solve medical mysteries through tracking and running ‘experiments’ to find out what is causing what. They asked me to be a beta tester, and I worked with them for a few months diligently tracking my food and symptoms. Through this exercise I learned of a few unexpected correlations: the burning appeared to be exacerbated by high-fructose ingredients like fruit and honey, and I could make the burning go away by eating garlic.

    Here’s the point: through self-tracking, I was able to pinpoint my burning triggers and thus I’ve been able to maintain complete control over my symptoms. When my symptoms flare, at least now I know why. There is nothing worse than experiencing a symptom and not understanding why it’s happening for where it’s coming from. And when your doctors are just as clueless, you feel pretty hopeless and helpless.

    I do want to admit that I don’t really enjoy the act of self-tracking. If I could be outfitted with sensors that could track the exact chemistry of my food, or the contractions of my gut, or the inflammation levels of my fingers and toes, then I would do it. But until that technology exists, we’re forced to do this in a bit more manual fashion. And at this point, for me, the benefits – at least of a short-term period of tracking to determine symptom causality/correlation – have far outweighed the burden.

    • Katie, you bring up a good point (several, actually) when you mention symptom tracking – and specifically for solving correlative mysteries like yours. You represent the subset of patients who are highly motivated to track – despite the fact that you don’t enjoy doing it.

      Many patients, however, lack either the will or the ability to do so – what Dr. Victor Montori and his Mayo Clinic-based team refer to as the “burden of treatment”. He writes, for example: “A patient’s education level, literacy, state of depression, pain, fatigue, social connectivity and supports, financial status – all of these affect a patient’s capacity to do the ‘work’ of being sick. This workload can simply exceed capacity to cope.”

      Expecting patients like this to start self-tracking adds to a workload few tech hypemeisters can even fathom.

      But the bigger question, as John points out so well, is what we’re going to actually DO with the data collected? Unless we’re planning to do something with it – or unless we happen to be QS data geeks who obsessively self-track just because we can – then why bother?

      • Carolyn, you might be touching on a challenging notion: creating effective feedback loops that drive behavior change.This is especially difficult in healthcare, particularly when the benefit to the *black box* can not be perceived by the *black box*.

        I would argue with John that tracking and moving the needle on cholesterol might be more difficult than tracking and addressing symptoms that are actually impinging upon the *black box’s* quality of life.

        Ultimately, every effort, especially those prescribed for people already burdened by disease/infirmity, must be simple, actionable and inclusive. If they’re not ALL three, they will not achieve longterm adoption.

  2. I love symptom tracking stuff, which has been around online since the 1990s. But this requires manual, active effort on the part of the patient, and for that reason, requires a fair amount of intrinsic motivation to do so.

    I think symptom tracking is great and can be potentially helpful in figuring out links that can’t otherwise be readily determined. Katie provides a perfect example of its beneficial use.

    But personal health records and symptom tracking require manual input of data — something I see as a continuing stumbling block for mainstream acceptance of any quantified self app.

    And as Carolyn alluded to, just like in personal health records, we have dozens of these apps (like Fitbit or 23andme) collecting health data for your benefit — yet they don’t talk to one another. There’s no simple vendor-neutral dashboard that could take feeds from any of these apps to show you your overall health. When we have one of these, I will be quite happy. :)

    • Hello John – when I read: “I love symptom tracking stuff”, I’m guessing that you’re not a patient living with a chronic diagnosis, or perhaps have not yet read ‘The Diabetic’s Paradox’ in The Atlantic about patients who say they “loathe” self-tracking – and it’s not just because the apps don’t talk to one another.

  3. In fact, 23andme can provide not just unconfirmed information but misleading information about the presence or absence of major disease due to the reliance on a certain set of SNPs.

    The interoperability issue is also a key one.

    Thanks, John.

  4. A few minutes after writing the above comment, I came across a Washington Post article from yesterday that reaffirms the problems with 23andme and other do-it-yourself genetic diagnostic offers, supporting John’s post and my comment.

    • Indeed. Companies that offer genetic testing are — in my opinion — just one step up from companies selling snake oil. Note this passage from the article, which amounts to an appeal to a testimonial:

      [Joanna Mountain, 23andMe’s senior director of research] says that customers often tell her stories about having medical conditions that went undiagnosed until they found out that they had a genetic propensity for the disease.

      I don’t expect scientists to be swayed by testimonials, and I don’t expect testimonials to be used a justification for a health product or service.

      While I believe most firms in this space are well-intentioned in their efforts, diseases are usually rarely simplistically genetically based (as the Washington Post article notes). They promise a lot, but its (mostly) based upon science that’s “not there yet.” Twenty years from now, who knows?

      But today, I think most people are just wasting their money to get these genetic tests.

      • I wonder if 23andMe has every queried their customer base as to their motivation for taking the test? We each come at this from our own narrative. I imagine there must be as many reasons for testing as their are people taking the test. I just wonder if there aren’t common drivers.

Leave a Reply