Ethan Zuckerman’s online home, since 2003
Currently Browsing: quantified self 2011

New sensors and the Quantified Self

This post is part of my liveblogged account of a conference. Two disclaimers: Liveblogging is hard, and I often get things wrong. If I did, please feel free to correct me via email or in the comments and I’ll make changes when appropriate. Second, the opinions expressed in these sorts of posts are those of the speakers, rather than mine.

Day two of the Quantified Self conference begins with a talk by Eric Boyd on New Sensors and the future of Self Tracking. Eric describes himself as “mostly a hacker”, someone who explores new technology and the capabilities they can provide us with.

He’s produced two very cool projects:

- Heart Spark, a pendant that flashes when your heart beats. This is less a quantified self project than a social communication one – what do we learn when we watch someone else’s heart rate increase as you’re talking.
- North Paw, a wearable compass gives you a sense of north by tingling on your ankle. This helps give you a perpetual sense of direction. Over time, he tells us, you lose the sense of vibration – you simply have a new sense.

Boyd’s talk is an overview of different sensor systems and what they might mean for personal tracking. He notes that not only have sensors gotten smaller and cheaper, but wireless and battery technologies have improved radically. As a result, it’s possible for companies like Green Goose to provide sensors in stickers. They’ve got accelerometers, low-power wifi and high density LiPo batteries that last three years. This means you can put a sticker on a pill bottle and tell whether you’ve taken your medicine, or put sensors on an asthma inhaler to measure where outbreaks are taking place, as Asthmapolis uses to map areas in a community where asthma attacks are common. As we think through the potential of these new sensors, there are lots of questions about open standards, and ensuring that this space remains interoperable.

But that’s not the focus of this talk – instead, it’s a tour of new sensors and capabilities.

Boyd begins with EMG – electromyography – the use of small skin-based electrodes to detect muscle activation. When a muscle contracts, it creates a small electric field. Neurosky, a brainwave monitoring headset, uses this sensor type. Amy Drill gave a talk at Quantified Self New York that showed off a pair of shorts with electrodes in it used to track and optimize the performance of olympic calibre athletes. While the system is currently expensive, they could easily come down in price, and would allow serious athletes and coaches to study the movement of individual muscle groups during activity.

Galvanic skin response sensors detect skin resistance, how much electricity flows across a gap on skin. Basically, this measures sweat levels. On a gross level, this is a way of sensing physical exertion. At lower levels, it can detect slight nervousness, agitation, and excitement. Paired with accelerometers and heart monitors, it might be possible to match mood information to physical activity.

Boyd is interested in glucometers, in part because he’s trying to debug a personal problem with low energy levels in the afternoon. Glucometers are pretty miserable at present, he notes – you spend $1 per test for a sensor that requires a drop of blood. What we hope for is a continuous monitor, but even bloodstream monitors at present need to be replaced every couple of days. The hope is that microneedles – patches of tiny needles with the texture of velcro – might be a solution for delivering vaccines through a skin patch, and eventually for continuous blood or fluid monitoring. Exciting, but these technologies are still in the lab.

Cameras are getting smaller and cheaper, and it’s worth asking whether a picture, traditionally worth a thousand words, could also be worth a thousand data points. Looxcie is a wearable camera that continually records. Press a button and the camera will store the previous 30 seconds. (I’ve wanted this functionality for years, and am thrilled someone’s actually built it.) Boyd talks about his dream camera-based tool, one that looks at faces of people who should be familiar to you and prompts you with their identity, perhaps via earbud. That’s a way off, but there are tools like Foodsnap that try to estimate your caloric load by allowing you to upload photos of your food. It’s hard to know how accurate these systems are – some are using Mechanical Turk to help with estimation. But even if you don’t look at the photos of food you’re eating, the act of photographing has a tendency to shape your diet.

Microphones are a sensor we tend to forget about. They’re cheap – often $2 – and can be used in interesting ways. One hacker put an air pillow in his bed with a microphone in it, and used the sound of airflow to measure his sleep movement and breathing. He got a huge amount of interesting data about sleep cycle from a sensor that was incredibly cheap. Boyd wonders what we might do with new sensors that detect ultrasonics, frequencies that humans can’t hear, but are used by bats and other animals. And he points to the Lena baby monitor, a $700 tool that listens to your child’s attempts at speech and tells you where your child is in the cycle of language development.

We’re seeing more sensors in our physical environment – the quantified world. Electricity monitors can actually tell us a lot about our personal behavior. Midnight bathroom breaks are visible as power fluctuations, as are the beginning and end of your time in bed. Automobiles are filled with an array of sensors, and using products like the Carchip Pro, which downloads automotive sensor data via the ODB2 port , you can access everything your car knows about itself, like tire pressure, speed, and engine RPM. Perhaps you could use this information as a way of detecting stress, if fast acceleration is a proxy for that behavior.

We’re seeing exciting challenges put on the table, like the Tricorder X Prize, recently launched with Qualcomm. It’s a $10m prize for a handheld device with multiple diagnostic capabilities. Boyd tells us that it’s really unlikely to be a separate, handheld unit like a Tricoder – it’s likely to be something strapped to the body. But we’re very early on in the idea, and it’s unclear what strategies might win out. The exciting nature of this moment in time is that there are lots of opportunities, both to play with sensors and to push the DIY aspect of the quantified self movement.


The business of Quantified Self

This post is part of my liveblogged account of a conference. Two disclaimers: Liveblogging is hard, and I often get things wrong. If I did, please feel free to correct me via email or in the comments and I’ll make changes when appropriate. Second, the opinions expressed in these sorts of posts are those of the speakers, rather than mine.

The closing session for the first day at the Quantified Self conference offers some glipses of the future of the movement. Paul Tarini of the Robert Wood Johnson Foundation has been instrumental in sponsoring some of QS’s early work (and the presence of several folks, myself included, ath the conference – thanks, Paul!) He introduces a new feature on Quantified Self’s website, a directory of products and services called The Complete QS Guide to Self Tracking, which features 400 tools at present.

One of the sexiest new tools is the Basis pulsetracer wristwatch, presented by Nadeem Kassam. He left the entertainment industry for health care and has been fascinated by the challenge of producing devices that are easy for many people to wear and enjoy, rather than existing for the obsessive folks who come to conferences like this one. The Basis device uses an optical blood flow sensor to measure bloodflow speed, which it can interpolate into a measure of hert rate. By combining this information with data from an accelerometer, galvanic response and temperature sensors, the “watch” can provide a very broad range of data on caloric burn, activity and sleep status.

The device isn’t the thing, Kassam tells us – it’s the way we present data to users. We need data that’s deep enough to be insightful, but simple and engaging. We need to be able to share it with other companies and systems. And we need to make it easy for users to wear, sync, learn and share, or we won’t find a way to extend personal tracking beyod the early adopter market.

The rest of the closing session focuses on precisely this challenge: turning personal tracking into a consumer product. Ben Rubin of Zeo (a sleep-tracking sensor), Jason Jacobs of Fitnesskeeper, and Brian Krejcarek of GreenGoose are wrestling with similar challenges. They’re building products based on their personal passions, but trying to sell to a broad audience. In the process, they’re learning a great deal about what might work – and not work – in bringing QS ideas to a wider audience.

Ben tells us that, while there are a few truly dedicated users who’ve used Zeo (a sensor that tracks sleep behavior) virtually every night since it’s been released, most users buy the product, use it intently for 3-4 weeks, and then fall off. What’s interesting is that they don’t stop using it entirely – the usage goes down, but six months after purchase, 70% were using it at least once a week. The hope is that by making a product convenient and easy to use, they’ll pick it up any time they have a sleep issue.

Jason, whose company makes software to track physical activity, has discovered that users who share their data on Facebook are more likely to stay engaged. So are users who integrate data from other tracking devices with data from their running or biking. And users who stop using the tools are often able to be lured back in by email prompts, that ask whether a user is “taking a break” and suggesting that they get back to tracking by setting and reaching a goal.

Brian, whose company manufactures inexpensive and small sensors that can record movement on objects like toothbrushes, pill bottles or water pipes, urges the audience to consider passive sensors rather than tools that require active data collection. The problem with a sensor you choose to use, he tells us, is that you end up with zeros for the days you didn’t participate. By making sensors pervasive, you can choose to ignore the data they transmit, but it’s there if and when you want it.

Ben – whose product is a sensor you need to choose to wear – allows that he’s a fan of ubiquitous, passive sensors. In the long term, we’ll have sensors in our beds, cars and phones… but it’s going to take a while. In the meantime, it makes sense to target sensors at problems people are having, like the need to get better sleep. And until there’s a richer ecosystem around these tools, a manufacturer may need to be highly vertically integrated. Zeo produces the physical sensor, the tools for visualizing sleep data, and the community that allows you to compare your sleep to that of others. It’s possible that, going forward, we’ll have a whole ecosystem of providers, but in the meantime, it makes sense to develop everything a user needs.

Gary Wolf, who’s moderating the session, asks all participants “What’s missing in the ecosystem?”

Ben suggests that stress is a market where there aren’t many good tools to analyze and understand a problem many individuals suffer from. In a more general sense, mass consumer awareness is still missing from the market as a whole. Brian suggests that personal tracking isn’t much fun – what’s missing is games that bring happiness to the process. Jason doesn’t believe anything in particular is missing from a QS ecosystem – instead, we just need more time to collect data and develop our tools.


Quantified Self – Location Tracking

This post is part of my liveblogged account of a conference. Two disclaimers: Liveblogging is hard, and I often get things wrong. If I did, please feel free to correct me via email or in the comments and I’ll make changes when appropriate. Second, the opinions expressed in these sorts of posts are those of the speakers, rather than mine.

Robin Barooah leads a session on location tracking at Quantified Self. He’s been developing a tool called Location Swap that is similar in functionality to Google Latitude – it allows you to track your location and share it with others. (He explains that he was working on the project well before Google launched the service.) He describes the ability to know where your partner or your friends are at all times as “an augmented sense”, an awareness that wasn’t possible before technological changes. The change, he offers, might be like the changes in behavior that come from having mobile phones and not being tied to landline phones.

The conversation quickly turns into a discussion of whether people really want to share their location, 24/7, with partners or others. Many of the people in the room would be willing to do so, and a few strongly object to the idea. One participant notes, “I used to live in Amsterdam, and culturally, no one closes their curtains. But there’s a strong cultural norm – you don’t look in the window.” Some of those who object to the idea of sharing location information aren’t worried about their movements, but don’t want their partner or friends feeling surveilled. Barooah admits that this information can be quite personal – he designed the tool initially to track and share his own behaviors, and ended up concluding that he wasn’t comfortable sharing that data.

Could we use location information to access sensor networks available in the physical world? What could we learn from tapping what’s being recorded around us? Josh Kaufmann recommends Asthmapolis, a system that maps the places in which asthma attacks are triggered, by attaching a GPS tag to inhalers, and sending location information to a server. What results is a map of areas in a city with particularly high levels of lung irritants, which might trigger protests against pollution.

There’s a great deal of concern about tracking systems that can’t be turned off. Mary Hodder talks about systems she’s built with telephone companies – it’s critical to have the ability to turn a system off. Tracking trucks for a trucking company is a legitimate activity while drivers are on duty, but the system needs to shut off when they’re off duty. On any of these systems, we need the ability to mediate who the information is shared with.

We talk about what tools people could use to track their data. Several people point out that using GPS continually on your phone tends to run down your batteries – Google Latitude may have done some smart thinking about this, and might be an option for some applications, even if the tool is designed primarily for sharing your location information with others. Mary points to some of the limitations of getting accurate tracking data – GPS is quite accurate, but not always available. On non-smartphones, AGPS – triangulation between towers – gets accuracy to a couple hundred feet. She notes that you can either collect your own data via applications or buy the data from carriers, which Loopt is evidently doing.

I asked people to talk about what they’re hoping to get out of tracking their location. Robin noted that location can be a proxy for behavior – if I’m in the park, I’m probably walking the dog. In the spirit of collecting as much data as possible, it seems silly not to collect this data, since it’s a form of passive sensing that’s perpetually available.

One participant is working on an app – tentatively named “Tripography” – that extrapolates what means of transportation you’re using based on your speed and calculates either calories burned (if you’re walking or biking) or CO2 emitted. The goal is to celebrate people for using low-carbon transport and to suggest alternatives. Another participant studies face to face social interaction and is interested in tracking location as a proxy for interaction. A third hopes to track location and correlate it to financial information – how much money do I spend per day when I’m in San Francisco, versus in Davis, CA? Josh suggests that we might learn from Mark Shepard, whose Sentient City Survival Kit includes an iPhone ap – Serendipitor – that will allow you to calculate a circuitous route between two locations in the hopes of having an unexpected encounter with something surprising and wonderful.


I was a little surprised that more individuals weren’t engaged with tracking their own location data. As at least one person in the session put it, “Well, I know where I am.” Of course, the whole premise of the quantitative self movement is that you often don’t know where you are until you have the data.

I’m fascinated by location tracking, because I suspect many of us inhabit much narrower physical spaces than we suspect or believe. (See my CHI talk on serendipity for more on this.) This session left me with the sense that the QS movement, in large part, is a personal health movement, at this point, rather than a personal data movement.


Targeted marketing at Quantified Self?

People at this conference are tracking aspects of their lives in far more detail than I currently am. I’m feeling very much like a late adopter and thinking I may need to run out and buy a Fitbit, a Zeo and start tracking my moods in terms of smiling panda bears.

But there are other forms of tracking that I’m just not ready to sign up for yet.

This ad, or a variant of it, appears over the urinals at the venue for the quantified self conference. I looked up BodyKey, of course, and discovered that they’re manufacturing an a tool for monitoring your health and education through your urine.

Perhaps there’s no such thing as too much data, but I’m pretty sure I’m not yet the target for this particular application…


Ignite Talks at Quantified Self

This post is part of my liveblogged account of a conference. Two disclaimers: Liveblogging is hard, and I often get things wrong. If I did, please feel free to correct me via email or in the comments and I’ll make changes when appropriate. Second, the opinions expressed in these sorts of posts are those of the speakers, rather than mine.

There’s a quick lunch break at Quantified Self followed by a series of five-minute Ignite talks. I find these virtually impossible to blog because of the speed, but here’s an attempt.

Rick Smolan wants us to know that big data is not big brother. The producer of projects like A Day in the Life of America, Smolan is focusing his next project on showing us the human side of big data and what we can learn from it. He mentions Ushahidi as an example of data being used to save lives. As we collect data from thousands of individuals, we can map the need for water or healthcare in parts of Haiti.

For a project that looks at a day in the life of big data, he’s going to set up 10 million human sensors, many of whom will download smartphone aps that track GPS location, steps taken, their mental state and ask questions each hour that ask people to help map their experiences and environment. This will be complemented by inputs from 1000 journalists in 50 countries. The goal is to understand how reflecting on data we collect can change our behavior, much like we change our driving by monitoring through the dashboard of a Prius. “These are reflections in a digital mirror – we can use big data to take the pulse of the planet.”


Misha David Chellam from Scanadu tells us that he didn’t know what a tricorder was until his 50-something business partner mentioned the Star Trek device as a metaphor for what they could build together. Misha is a geek twentysomething who likes cool devices, and his partner doesn’t want to die, so together, they’re trying to build the medical tricorder.

There’s lots of health data available these days, from self-tracking devices like the Fitbit and the Zeo, from “macro-scanning” tools like full body scans and genomic analysis, like 23 and me, and we now have access to digital, “nomadic” health record systems like Practice Fusion, Google Health and Microsoft’s Health Vault. We could add to this “sequencing human lifestyles”, data that helps us understand behaviors on a population level.

The next step is interpreting this data. For starters, we can try to do interpretation using doctors in a Mechanical Turk fashion, perhaps using doctors who are solely in private practice and carrying a lower patient load. In the long run, we might do AI – and Watson’s victory on Jeopardy is an inspiration.

The tricorder is the metaphor because it does so many things… and the contemporary tricorder is the mobile phone. It’s got a vast number of sensors that are helpful to us, and we can add to it with interfaces like microfluidics readers. The vision behind Scanadu is developing a strategy that can win an XPrize, focused on building a tricoder that can evaluate a patient better than a board of physicians. Scanadu is taking first steps to this, collecting blood from alpha users and using Wolfram Alpha for contextualizing this data. It’s a first step, but they’ll have lots more to work with if they can partner with tracking device developers.


Alan Gale of Bio-Logic Health is interested in life extension through food and supplement tracking. In the past, Gale tells us, our methods for life extension were pretty weak: mummification, drinking blood, freezing Ted Williams’s head. Current approaches, advocated by people like Aubrey De Grey and Ray Kurzweil either focus on future technologies, or on new developments in regenerative medicine and hormone replacement.

At present, the best preventative techniques we know about are caloric restriction and supplements, a regimen that requires massive lifestyle changes. You have to take hundreds of supplements, some of which can be toxic in high doses. Managing this process requires lots of careful adherence and tracking.

Gale views the human body from an engineering point of view. It’s composed of subsystems, each with inputs and outptus. Systems are regulated both via feedback mechanisms, or through our conscious intervention. When it gets cold, we can shiver (activating endocrine and muscular systems) or we can put on a sweater. The same is true with food. With mobile tools, we can track inputs like what we eat, and outputs like our blood tests. Over time we can build a model of feedback mechanisms, guiding people towards their deficiencies and meeting their goals.


Sarah Gray tells a story about tracking her mood that starts, as most good stories do, with a boy. The boy lived in a different city, and she found herself unable to decide whether she should move to be with him, continue a long distance relationship or move on. So she built a website that allowed her to track her feelings. Over a few months, she rated her mood from 1-5 and looked for patterns. After a few months, her understanding of the situation was much clearer, and she decided to separate from the guy in question.

The app she built is the root of MercuryApp, a mood tracking website designed for easy use with smartphones. She suggests that it works because it encourages a ritual where we track every day, encourages reflection, where we stop and think about what’s going on, and helps us find a story, an arc, to our behavior.

She offers examples of individuals using the app:

- Sebastian, an pathological optimist, who thinks situations are always going to get better. After discovering that he was unhappy at work, week after week, he decided to quit his job and move back to Spain with his wife. “You can write off one sad panda, but not a string of them.”

- Dave, who manages an embedded software team. The team members use the tool to track their morale, and Dave has a real-time health check on the mood of the team.

The goal is to merge hard and soft data to help individuals become happier. Answering the question, “When are you happiest?” requires both quantifiable data and the data of your gut.


Marcy Swenson and Dale Larson offer a skit to explain what agile development might teach us about personal tracking. Dale’s worried about falling asleep during a session this afternoon. So he plans an experiment to use his Zeo, measure his sleep against caffeine consumption, mood, food and exercise data, then graph it all and engage in multivariate analysis to solve the problem!

Marcy observes that Dale seems to be more focused on data than on solving the problem. If we learned from software development, we might try weekly sprints, information radiators and a tight build/measure/learn cycle which might let us figure out what we really needed to know before investing months in a particular process. They’ve expanded on some of these thoughts at startuphappiness.com


Ron Gutman wants us to know about the untapped power of the smile. He’s a serious runner, and discovered that when he hits the wall in a long run (75 minutes in!), he often feels better when he smiles. He began tracking the data closely and discovered it was an unambiguous correlation for him. So he began a wide-ranging study of the power of the smile.

A 30 year longitudinal study tracked the relationship between people’s happiness (on a test of well-being) and success of their marriages and their high school yearbook photos. Based on people’s smiles, researchers could make very accurate predictions of the future of these students. Another study looked at pre-1950 baseball cards. The span of a player’s smile could predict span of the player’s life – bigger smiles predicts longevity.

Less than 15% of people smile less than 5 times a day. At the same time less than 1/3rd smile more than 20 times. It’s certainly possible to smile more: children can smile up to 400 times a day. It’s possible that smiling can, in and of iself, make us feel better. Charles Darwin speculated, “Even the simulation of an emotion tends to arouse it in our minds.”

Gutman tells us that one smile can create the same brain stimulation as 2000 bars of chocolate (with fewer calories.) If you smile, others see your smile and feel good. In turn, they smile and you feel good. He closes with a quote from Mother Theresa: “I will never understand all the good that a simple smile can accomplish.”


Sean Ahrens has Crohn’s disease, an inflammation in the digestive track caused from a disregulated immune system. He’s coped with the disease for 13 years, and recently decided to take some worms. He took pig whipworm, and a friend took human hookworm. It’s not a cheap thing to do – he spent $3000 to purchase worm eggs from Germany and Thailand, which he took every two weeks for five months. The eggs appeared, under a $12 microscope, to be worm eggs. And Ahrens monitored his symptons – gut pain and bowel movements – closely for the months he took the eggs and months afterwards.

It wasn’t a very successful experiment, both in that his symptoms didn’t get much better after taking the eggs, and that he can’t definitively say whether the eggs failed. First, he didn’t have much baseline data. Second, there were other changes in the time he tracked – a change of diet, other medications, and stress from participating in Y-Combinator. He ended up concluding that he didn’t have enough background in math to figure out causality in the data. The talk ends up being a cautionary tale about getting baselines and controlling experiments… which can be hard to do when you want pain to go away. In the meantime, Ahrens is working on a company, Crohnology, that’s a supportive social network for people with the disease.


Tina Park is a designer for Johnson and Johnson who worked on Project Health Design, a effort from the Robert Wood Johnson Foundation’s Pioneers Program to help teenagers with chronic conditions transition from pediatric to adult health care. This tends to be a difficult transition: teenagers go through lots of life changes, teens forget medications, and they can get sick and die.

It’s possible, she argues, to track teen’s moods through texts. And since teens identify health through mood, it’s possible to identify moments where teens may be unwell by reading their text messages. Many teens send hundreds of messages a day. Her project graphed the intensity of message sending on timelines – she shows us a visualization of data of a teen’s data for six months. You can tell when she’s asleep based on the flat periods in the timelines. And you can see what words are common at different times in the series, which can help a teen see what she was talking about and, perhaps, what stressors are happening in her life.

The benefit – this is data that already exists – perhaps we can get insights on mood without collecting any additional data.


« Previous Entries Next Entries »

Powered by WordPress | Designed by Elegant Themes