by Gary Wolf
Thank you to the many, many people who sent us private and public feedback on this post. In response to your comments and some revised thinking among our collaborators, I’ve made one substantive change. A recently published paper by Nils Heyen uses “personal science” rather than “everyday science” to describe the kind of self-research discussed below. His paper is convincing, and I’ve replaced everyday science with personal science throughout the essay below. See: From self-tracking to self-expertise: The production of self-related knowledge by doing personal science.
This device has one button. When you press the button, it records the time. It isn’t necessary to set the clock, which is always accurate. Connect the device to your computer and it opens as a drive. The record of your button presses is there as a file on the drive. This device is useful because it provides a convenient way to record accurate observations in daily life. Unlike a mobile app, it can be used discreetly, privately, and won’t go obsolete. Self-trackers and clinical researchers are already using it today to learn about triggers of PTSD, restless leg syndrome, and allergy.
The 1 button tracker was created by Thomas Blomseth Christiansen and Jakob Eg Larsen at Totti Labs in Copenhagen. Anybody with experience doing empirical research can easily imagine a hundred different ways to use it. But the reason I start with this picture is to pose a more systematic question about how we acquire tools for health discovery: Why, after a decade that has seen over US$30 billion invested in digital health, has nothing like this simple tool even been contemplated, much less delivered as a product, by the commercial tech industry?
I’m currently using the 1 Button Tracker to learn about my tremor. My tremor condition, known as “essential tremor,” is shared by 2%-5% of adults over 50 years old. It is a common, progressive, and very annoying condition, for which there is no cure. I know that my tremor varies, and I want to understand what influences it so that I can reduce the annoyance and maybe even slow the progression. So when my tremor comes on, I press the tracker. With minimal effort, I now have a record of my symptoms, accurate observations that are the first step to understanding.
If you know anything about trends in consumer electronics, you may find it strange that I’m trying to make accurate observations my tremor, which is a distinctly physical phenomenon, by pressing a button when I make a mental note that it’s especially bad. My tremor could also be detected with an accelerometer, and there are accelerometers in nearly all of the wearables that have ever been sold, as well as in billions of smart phones. Why not use one of these?
The reason is there is no tremor tracking app that takes advantage of the accelerometers in wearables or phone. The reason there’s no tremor tracker app, is that there aren’t any generally applicable interventions for essential tremor short of brain surgery. (Mine is not serious enough to justify brain surgery, and although I know the condition is progressive I hope to avoid this intervention.) No common interventions for my kind of tremor means no business model. No business model means no investment. It’s all very rational: Why assess a condition when there aren’t products you can sell to treat it?
And yet there is still value to me in having a better understanding of my tremor. How fast is it progressing? Is there anything I’m doing in my daily life that makes it worse or better?
Fortunately, there’s another way forward. With an excellent free app called Physics Toolbox Sensor Suite by Chrystian and Rebecca Vieyra, I can get a readout of the accelerometer data from my iPhone. To track my tremor, I balance it on my thumb, where the tremor is especially pronounced. Thanks to assistance from Mad Price Ball and Beau Gunderson, I upload my sensor data to Open Humans and get a tremor score. Mad and Beau and I put this workflow together in a few hours over a few days.
My tremor tracker is not a finished product. It isn’t really a product at all. Instead, it’s one step in an ongoing process of learning and self-discovery. The fact that within a few weeks of thinking about it I had a workflow that gave me accurate physiological measurement of tremor is also a hint about a new way to use shared tools and peer support to dramatically accelerate the pace of health discovery.
Making new discoveries about health is normally considered a job for professionals. The gold standard for health research is the large, anonymized, randomized trial, organized from the top down by experts. It has proven it’s power, but it remains silent on a vast number of questions that are vitally important to health. After all, everybody has questions clinical research will never attempt to answer. No matter how many good clinical trials the professional scientists complete, we still want to know: What is true for me?
Over the last decade, in volunteer meetups in over 100 cities in thirty countries, and through a dozen international conferences in the United States and Europe, people in the Quantified Self community have shared the tools and methods they’ve used to investigate their own questions with their own data. With my collaborator Steven Jonas and other colleagues, I’ve transcribed, tagged, and published hundreds of these projects.
Most of these people doing these self-tracking projects have some scientific or technical training, and they are comfortable using DIY tools. However, it’s important to notice that while their skills may be exceptional, their questions are common. Self-trackers in the Quantified Self community want to know how to cope with a disease or an injury, how to improve mental health, and how to make good decisions about their own medical care.
People exploring their own questions with their own data face significant challenges; and yet, many of these challenges have been solved already, somewhere, by somebody. The power of the Quantified Self community lies in the connections it encourages among people who can learn from one another. Our challenge now is to scale this knowledge sharing ecosystem so that more people can benefit. I envision a world with as many new, consequential health discoveries as there are articles in Wikipedia — and even more.
When I say that we can support tens of millions of new health discoveries, you’ll be right to ask: What exactly do I mean by discoveries? Let me be clear that I’m not talking about disciplinary contributions to research science. From professional health science we want generalizable knowledge that leads to new healthcare products, policies, and standards of care. Discoveries that apply to one person are uninteresting. However, when it comes to exploring our own health questions, we’re happy if we can learn something consequential about our own health, even if nobody else ever asks exactly the same question. When it comes to our own individual questions, the empirical methods of science — accurate observation, rational analysis of data, creative thinking about how to test your ideas — are highly useful. We call this “personal science” to distinguish it from academic and clinical research.
Personal science today may seem like the geekiest of practices, as fiddly to apply and as far out on the mainstream adoption curve as editing in a wiki in the 1990s or using a handheld computer in the 1980s. But the emergence of open tools and shared workflows is the first step in broadening access to personal science to everybody.
If thinking for ourselves is so useful, why hasn’t the tech industry done it for us already?
I put the question this way as a joke, but it expresses a curious fact. Countless wearables companies are pitching themselves to healthcare, and most of them offer carefully calibrated physiological observations as part of the package. We carry a lab in our pocket, and even on our wrist. Can’t we find our answers there?
Unfortunately, the requirements of personal science are profoundly out of synch with the healthcare industry in general and digital health specifically. The answers offered by digital health are only those that meet the demands of the market: potentially profitable solutions to widespread health issues, as these are framed and understood, not by the individuals affected, but by the system’s gatekeepers: providers, payers, and regulators. Talk honestly with any digital health expert — especially one who has tried to change the system through starting a company that addresses these failures––and you’ll get an earful about misaligned incentives and crippling inertia. Most importantly, the focus of digital health is not on observations but on interventions. Delivering interventions into the healthcare market is complex and risky, luring many startups into building themselves a competitive “moat” with privacy abusing surveillance technology, black box algorithms and centralized data control.
The practice of personal science requires tools designed with different principles altogether. Because personal science is a personal affair, our tools need to store data privately, with all sharing options under full user control. Because complexity is a typical cause of failure, self-tracking tools need to be simple. Because success in self-research often hinges on developing a unique protocol for making observations — what’s observed, when it’s observed, and how often — our tools need to be flexible and self-governed. Because we need to be able to copy and extend techniques others have used, our tools have to be modular. And because self-research may go in unexpected directions, and may go on for many years, our tools need to be open and inspectable at all levels so they have resilience over time and can be thoroughly examined and understood.
Some tools that meet these requirements exist. Many still need to be created. Together, they comprise what I call the open stack for self tracking. The open stack isn’t an existing set of off-the-shelf components, but rather an expanding set of tools that meet the design requirements of Personal science: Private. Simple. Self-Governed. Open.
In today’s Quantified Self community, this open framework is more or less hand operated. People come to meetings, post online and support each other in private Facebook groups and patient communities. They exchange tools and ideas, present their discoveries, copy workflows, get feedback, share protocols. But the continued development of open self-tracking tools will allow more scalable and online support to evolve.
We live in an age of participation. Technology that makes it easy for people to make, discover, and share things has fundamentally reshaped older systems of mass production in many domains. Why not in science? The diversity of needs and questions people have, and the diversity of skills and experiences required to address them, makes this a natural job for participatory, peer-to-peer production. Healthcare may be slow in helping. But we can help each other.
Supporting open tools and practices can have positive, large scale, systematic effects. The open stack for self-tracking is needed for our own individual projects. But also has the power to accelerate academic and clinical research by easing the tension between privacy and sharing that has long stymied collective approaches. Instead of Hoovering up everything about you into an insecure, fragile, big data infrastructure, an open stack approach brings the infrastructure to your data, and you retain control. This makes shared discovery possible without putting people at risk.
For instance, in personal science, we typically want to record when something happens, where it happens, and what else is occurring around the same times and places. We need measurements, of course, but we also need a way to line them up, to see them on a graph or as a numerical score, to compare yesterday and today, to make an average over time, or to watch two different kinds of measurements rise and fall together. Fortunately, it is now possible to run many of these common analytical operations in a web browser without sharing your data, using open source tools such as Jupyter notebooks. The next step is to harness the power of participatory peer-to-peer production and to make it easier to copy, apply, and extend each others’ notebooks. Imagine interactive analytical notebooks on tens of thousands of topics reading data from any file that a user has in possession — without demanding that they share their data. These are the foundations of a scalable system of support for everyday science for anybody.
We’re also beginning to see a “demonetization” of the observation layer of the self-tracking stack; that is, the sensors, hardware, and software for collecting our own data. Freeing the observation layer has the potential to dramatically accelerate discovery, both individually and collectively. Every digital health lab has graduate students dedicated to maintaining fiddly workflows for extracting data from commercial wearables; that is, where they aren’t paying excessively for specialized research instrumentation containing cheap sensors. This is wasteful. When commodity sensors, simple event detection, and accessible methods of everyday reasoning are freed from tight integration into commercial products, investment and innovation can cleanly focus on experiments and interventions, where expensive treatment evidence is required. Meanwhile, we all get more freedom to think for ourselves.
To get you started in thinking about what a world of open tools for self-tracking will be like, take a look at the sensors on the Card10, an open source instrument given out as the badge for participants in the Chaos Communication Camp near Berlin last summer. The Card10 comes with sensors for humidity, air pressure, temperature and air quality as well as a virtual gyroscope (accelerometer + magnetometer), ECG & heart rate sensor. While there is still a lot of work involved in making open instruments accessible to more people, the question of what kinds of measurements can and should be provided by the open stack has its first set of answers already.
Of course, much of the work to be done is not technical. Science teachers at both the primary and secondary level will play a crucial role in expanding access to personal science, as new tools can help students begin empirical adventures into the world of their own experience. Training for health care allied professionals is also badly needed, since many people exploring personal questions with personal science are dependent on healthcare for important resources, such as assessments, drugs, medical implants, and surgeries. Through collaboration with Dr. Martijn de Groot’s program training health care professionals at the QS Institute in the Netherlands, we’ve seen how nurses and physical therapists can make personal science part of their toolkit, deepening their commitment to make practice more like research, and research more like practice.
We have a long way to go, but we can build on this progress to address the health and civic emergencies we face in the 21st century. When I think about a world that allows everybody the tools and support they need to explore their own questions empirically —my world of 10 million discoveries — I find myself inspired not only by the potential content of these discoveries but also by the positive cultural challenge they pose in a world where a global crisis of authority has damaged civic trust. The ability to make our own discoveries fundamentally alters the power dynamics of healthcare and public health. Meanwhile, broad participation in reasoning about ourselves strengthens our mutual respect for one another as rational beings, each of us with a legitimate interest in figuring out our own future. This mutual respect is fundamental to democracy, indispensable to long term survival — and far from secure.
If this topic is new to you, I hope I’ve prompted you to think about how the power of personal science could help you explore your own questions. And if you’re already involved in self-research or are making self-tracking tools that could be useful to others, please let us know about your projects. Our hashtag for this essay is #10milliondiscoveries, and I’ll keep an eye out for your replies.