"Introducing ResearchKit. A software framework designed to benefit medical research," began a new section during yesterday's extended Apple event, which largely focused on easy-to-understand product announcements like a new MacBook laptop and the new Apple Watch. "New apps for Parkinson's, diabetes, asthma, heart health, breast cancer—this could be HUGE. No evident profit motive—Apple at its best," wrote former New York Times tech columnist David Pogue on Twitter, lost in reverie for the company.
ResearchKit is a theoretically benevolent mirror image of the various data collection scandals that have come to define the tech industry over the past few years: yes, massive corporations (I am including the United States government under the umbrella of "massive corporation," because I am a cool edgy blogger) are collecting and using massive amounts of your most personal and private data in ways you can't imagine. But what if that was all for a good cause? I tried out one of the new ResearchKit apps, and spoke to one of that app's creators, to see if I could figure out how it all works—and maybe, from my couch, in my underwear, cure Parkinson's.
iPhones, and, soon, Apple Watches, will collect an astounding amount of data through various hardware and software. Your iPhone tracks how many steps you take and where you go, and it is not technically possible to stop it from doing that.
To get useful data out of Apple's health apps, you have to add even more information to them: height, weight, age, gender, even blood type. (The Watch will add heart rate monitoring to that, which can be extrapolated into much more data, like the specific type of exercise you're doing and how many calories that burns.)
I downloaded "Parkinson mPower study app," a brand-new, free app from the nonprofit Sage Bionetworks working in concert with various research and medical centers. It promises the "world's largest and most comprehensive study of this disease."
The app begins by forcing you through a "Consent" process, in which it, very succinctly and honestly, tells you exactly how much data it's going to need from you. (A lot.) It also warned me that "Participating in this study could generate a wide range of emotions." (It didn't, really.)
The tests struck me as incredibly vague
I chose "Yes" on every option. I gave Sage Bionetworks information about my gender, marital status, employment status, age, and whether I've had Parkinson's-related surgery. I informed the app that I do not smoke (there was no option for "maybe two cigarettes a month while drinking" so I rounded down). I gave the app estimations of when I usually go to sleep and when I wake up. It asked whether I have ever had something called "Deep Brain Stimulation," capital letters sic. I felt a little sheepish answering no, as if I had never really plumbed the depths of my mind. It asked for permissions to use various sensors, and I said yes to all of them: now the app knows my location, has access to my microphone, knows how many steps I take thanks to the motion activity sensor, and, for good measure, I gave it permission to bombard me with notifications. Then it asked for my name, and my signature. When you wake your phone and switch back to the mPower app, it asks you to authenticate with your Touch ID, the thumbprint scanner, so I suppose it has my thumbprint, too. (You can use a PIN for this, but I didn't.)
This is a very long process, and at the end, it gives you a quiz to make sure you understood, for example, that copious amounts of your personal data will be used by various researchers, scientists, and engineers, and that while Sage Bionetworks will do their damndest to make sure it's all anonymous, there's totally a chance that they'll get hacked and your data will end up on KickassTorrents.
The app has a few very basic game-type tests available on your first day of testing. I performed a memory test, which was like a minimalist version of the memory game Simon; a thing where I said "ahhhhhh" for as long as I could; and a thing where I tapped two fingers as fast as I could. The whole thing took maybe four minutes. Afterwards I was presented with a chart showing me how I did, which I already knew because I just did it, and also other data that I had allowed the app to know, like how many steps I had taken that day.
Unsure about exactly what I had done, I called Dr. Stephen H. Friend, the president of Sage Bionetworks. He told me that the tests on the app are all versions of tests Parkinson's patients perform regularly; some are more effective than the traditional non-app versions, some less.
"The tapping test is a way of doing something that's equivalent," he said. "Someone would be asked to pinch their fingers together instead of tapping." The memory test is inferior; usually a patient would go through what Friend calls "a 45-minute battery" of memory tests at the doctor's office.
The voice test is more detailed, however. "In the voice there are more than 100 dimensions that we can track, and we end up getting very accurate information on the rigidity [of the vocal cords]," he said.
The tests struck me as incredibly vague. There are lots of existing medical apps that seem similar but aren't, like the WebMD app for self-diagnosing; Cell Slider, a game in which users identify cancerous cells on images; or any of the various apps that help patients communicate directly with their doctors.
The whole point is to cast as wide a net as possible: get users, and get data, and have researchers take a crack at combing through it all
Sage Bionetworks isn't seeking to solve one particular problem, which Friend says is intentional. "We could have done this as a narrow study to show the power of numbers on a particular question," he said.
Instead, the whole point is to cast as wide a net as possible: get users, and get data, and have researchers take a crack at combing through it all to figure out exactly what an open-source, crowdsource research app can be used for. As for right now, "I don't think we know the right questions to ask," he said. And nobody really knows what this data might illuminate. All we know is that researchers don't currently have access to it.
There's a great deal they're unsure of, actually. There's not much of a gamification aspect to the app; RunKeeper this is not. But, admitted Friend, the app is really at its most useful when each user continues to provide data every day, to see exactly how things change over time in response to different environmental factors. So how do you get people to keep playing an extremely spare version of Simon?
"I think no one has broken the code on how to get longterm participation by patients in health apps," he said. But, he thinks, "as this evolves, the sense of doing something for others, not for yourself, is a stronger pull."
That strikes me as overly optimistic, but I don't have Parkinson's, nor do I know anybody who does. Maybe if I did I'd feel more of a push to participate.
We spoke about privacy a bit. Privacy is a huge concern for mPower, as it should be for any app in the ResearchKit category; how many apps tell you upfront that they'll do their best, but sometimes apps get hacked, in which case you're screwed?
mPower anonymizes a user's identity from that user's data, says Friend: "A researcher would love to go back and say, 'well, patient 321, look what happened here, here, and here, I'd love to contact them,' and we said when we built this that we can't do that, because someone might try to reverse engineer it, or the government would say they need to know who that person is."
So why is the app even asking for my name? What's the point, if it's going to be anonymized anyway? Friend says they request full names and email addresses to make people feel more comfortable, because a doctor would ask for those things, in addition to functioning as a sort of Captcha to make sure someone using the app is a real person.
Big Data comes with a set of pitfalls that medical researchers have never really had to deal with before
It is standard practice to use a consent form in research studies; this is considered a tentpole of medical ethics. And it's a good example of the struggle that the medical world is going to have to deal with as it moves into the brave new world of Big Data: ResearchKit is certainly the wave of the future, or at least a wave of a future, but Big Data comes with a set of pitfalls that medical researchers have never really had to deal with before.
There are ethics related to mass collection of data, and ethics related to collecting medical data for research surveys, and those ethics are, suddenly, directly at odds. For researchers to be comfortable (and, they assume, for subjects to feel comfortable), they need an identity authenticated. But for contributors to Big Data—in other words, smartphone users—to feel comfortable, they need to completely avoid the question of identity at all. That's the central push-and-pull of turning the field of medicine into Google Maps or Kickstarter, and mPower hasn't figured it out. If I weren't writing this article, I would have immediately uninstalled the app as soon as it asked me for my real name.
ResearchKit, which at launch also includes an app to help educate and monitor asthma patients and an app similar to mPower that's designed to connect risk factors in daily life with heart disease, is a big idea, certainly. Let's take all that data that companies like Apple, Facebook, and Google are using to sell you stuff (or to sell to people who want to sell you stuff more effectively) and use it for something objectively good. But given that one of the central societal concerns of this decade is privacy, who has your information and what are they doing with it—does it even matter if the goal is good? After all, the NSA claimed the same thing.