Lion or Bear? Self-Tracking and Social Identity

Gary Wolf

From Ethan Zuckerman's always interesting blog, My heart's in Accra, comes this story of a prototype of a social tracking device that helps teenagers notice patterns in their social behavior - and also alerts their counselors:

Ennea, a project from students at the Eindhoven University of Technology is one of the cooler things I've seen in a long time, developed during a six week design class. The students focused on an interesting problem - the problems incoming Dutch high-school students have in building socialization skills. The Dutch education system doesn't have middle schools, so students go directly from an elementary school to high school, a transition that can be difficult and stressful. Schools assign "tutors" to groups of pupils, and they meet for an hour a week to work on socialization skills. The designers talked with tutors and realized they had very little information about how their students were doing, and designed a fascinating social tool that works as a very clever form of surveillance and behavior tracking.

The designers produced a set of small, cute, wireless-aware objects that students carried with them for a few weeks. The objects measured interactions between children, timing the interactions each child had, and whether they were with individuals or groups. This information allows the designers to describe each child's interactions in a two-dimensional matrix based on interaction diversity and intensity. (Meet a lot of people and you're more diverse. Spend a long time with a person, and it's more intense.)

Rather than scoring the children on good or bad types of interaction, the device characterizes a user as one of nine animals: Lions are very diverse and very intense in their interactions. Their opposites are Polar Bears, who interact infrequently and briefly. Users can change roles over time - the device vibrates when your state changes, but you can only see what role you've taken on by "mating" your device with another person's device, giving the opportunity for conversation and interaction. For "complementary" roles, the animal icons will glow gold.

The idea is similar to an experiment I saw Nathan Eagle describe about last year at ETech: a hundred MIT students were given free use of smart phones in exchange for participating in an experiment that used low power Bluetooth transmissions to track when students were together and for how long. But the twist here makes it much more interesting, for the Ennea designers have created a feedback system designed to make the students' patterns visible to themselves. They are also being tracked, of course, something that I would certainly have resisted passionately when I was a student. As Ethan points out, the dark spirit of total surveillance haunts technologies like these.

Excise the authorities (we're allowed to dream), and the game is wonderful. Having the device signal when you change your pattern, and undergo a metamorphosis from one animal to another is brilliant. These moments are hard for us to notice in the near term; often a long time passes before we are able to look back and say: hey, I was changing.

First Personal Genome User Group

Kevin Kelly

Last Tuesday 23andMe, the genome sequencing service, hosted the first meeting between its customers. Like Navigenics and deCode, 23andMe will sequence half a million "snips" from your own DNA to give you glimpse of your personal genetics. What you get for your $1,000 payment is lots of numbers and strings of letters, and a little bit of information of how these correspond to health studies and ancestry research. So far this is a very nerdy hobby, requiring a love of data and great patience with crude tools, bugs, and beta everything. 

The idea behind gathering early enthusiasts face to face was to increase the exchange of information, methods, motivations in a way that Web 2.0 can't do. There are ethical issues, social uncertainty, incredible technical complexity, and rapid advancement in this field; having a person in front of you who might know more than you is a great way to learn. 23andMe cleverly calls this initiative to harness customer-generated value 23andWe.

As a paying customer I attended this informal gen-con. I think of this event as the first Personal Genome User Group. We gathered at 23andMe world headquarters in Mountain View, CA, but I think we were outnumbered by the ever helpful staff on hand to answer questions. I was extremely curious about the other customers' experience, and what they have learned by sequencing their DNA. I don't want to speak for them, but here is what I have learned by messing around in personal quantified genomics in the last six months:

1) There is far less known about proven genetic diseases that I thought. The list of illness with established origins in genetic mutations is very short, and even then the genetic factors operate statistically. The presence of a particular gene will raise or lower the percentage risk for a health issue, at the most. So one thing I learned from having my DNA sequenced is that very few diseases have a single point genetic basis, which means that gene sequences will be just one factor in a very complex brew of influences.  Personal genetic knowledge is sort of like knowing your blood type: Important to know but far from sufficient.

2) Sequencing is not just about health (and one of the reasons it should not be hijacked by the doctor squad). Your DNA can reveal much about your deep genetic past. Combined with research around the world into ethnic locations and migrations, your sequence can map out the movements of your distant ancestors in pre-history. I was surprised by how interested I became in these movements, both on my maternal as well as paternal side.  I had zero interest in genealogy, and could not care less about my great-great-grandparents and their cousins. But I did find myself surprised by the ethnic pathways of my ancestors, because the identity I normally cling to is obviously (duh!) just the most recent one in a very long line. Naming myself after only one data point and ignoring all the others seems arbitrary. Many of other users at this meeting also expressed to me this pleasant surprise of discovering their interest in their deep ancestry via their sequences.

3) 23andWe is the perfect turn of phrase for another thing I learned.  I have been surprised at how fast and how eager users have been to share their  genetic data. We've been conditioned by anxious media reports to believe that people want to hoard their very personal genetic profile, in fear of what would happen if governments, corporations, insurance companies and the neighbors were to see it. But in fact like a lot of other things that have made it online, genetic information only increases in value when shared. Experts thought only a fringe minority would dare share their genes, but swapping genetic info will mostly likely be the norm for a generation that shares everything else. Sharing your genetic info with family members, relatives, and even apparent strangers (who must be related somehow) is exciting, and certainly educational.

Gary Wolf and I have discussed starting a Bay Area Quantified Self User Group with a similar mission: to mutually assist those folks interested in inventing and mastering the tools of self-knowledge, including genetic sequencing. Write me if you are interested.

From Self-Observation to Medicine

Kevin Kelly

The art of constant self-awareness and self-experimentation is essential to the habit of self-metrics. Occasionally a trained scientist can take a small signal from their own life and turn it into a falsifiable result. I found the following note of such self-observation on the website for The People's Pharmacy. This center for patient submitted alternative remedies has been run by Joe and Terry Graedon since the 1970s.  Among the many recent examples of self-awareness leading to quantifiable medical results they list this story:

This is one of the most bizarre discoveries we have ever heard about.

A scientist noticed that when he got up out of his chair after watching a football game, his arthritis pain was greatly diminished.

During the course of the game he had consumed a six-pack of diet soda containing aspartame. Putting two and two together, he thought this artificial sweetener might have contributed to his relief.

He organized a placebo-controlled trial involving aspartame (aka Equal, NutraSweet) and confirmed that doses of 76 to 152 mg did indeed provide pain relief, roughly comparable to anti-inflammatory agents.

This research was published in the very respectable scientific journal, Clinical Pharmacology and Therapeutics.*

*Edmundson, A. B., and C. V. Manion. "Treatment of Osteoarthritis with Aspartame." Clin. Pharmacol. Ther. 1998; 63:580-593.

Detailed quantifiable self-observation has a new handle. It is called ODL or Observations of Daily Living. The idea is that if you monitor your body in your daily life over time you'll have more than just a snapshot of your health, you'll have baselines and long-term trends.  As Project Health Design suggests:

..The importance of observations of daily living (ODL) [is] moving toward next-generation personal health records and health management. 

Things like:  Does my chronic pain spike when the temperature dips below a certain threshold?  What effects might a particularly stressful month, with long hours at work and marginal sleep, have on my eating and activity behaviors, and hence my diabetes?  Can the fact that a 17-year-old with a chronic illness is regularly self-reporting his or her mood to be bad or sad play a role in the self-management of his or her disease?  And, if today’s pollen counts are really high, can my PHR device send me an alert in the morning to remember my inhaler, and then delete that point-in-time data capture because it may not be useful if conditions change tomorrow?

Trixie Tracker: Data-driven Parenting

Kevin Kelly

The Quantified Self is primarily about self-monitoring, and not about monitoring others. But your baby is close to the self, so there may be some technology in baby monitoring to be of use to adults.

Trixie Tracker tracks and displays the activity patterns of babies. As they claim on their website:

"Uncover patterns in your baby's sleep rhythms and daily activity. Develop a good sleep schedule with helpful charts. Share online with family and friends."

Using this software parents can track what goes into baby, and when; what comes out, when; when baby sleeps, when baby wakes, and any other activity you want to collect data for.

This is a primitive version 1.0 of tracking tools because you need to manually enter all data. The tool provides a web-based fancy spreadsheet with cute charts. You provide the data entry.  It has an iPhone input option, too, which could make a difference. What you can take away -- particularly if you are willing to share your baby' data -- is some sophisticated analysis of say baby's sleep probability.

Trixie Tracker is part of a larger idea called data-driven parenting, which I suspect has a small following right now, because who wants to spend their lives inputing data? But once all these ubiquitous devices collect data for us, crunching your kids' day later in the evening after they go to bed may be the new parental chore.

Annals of Self-Experiment - Seth Roberts is His Own Mouse

Gary Wolf

I'm becoming a devoted fan of Seth Roberts, one of the great champion of self-experimentation. Roberts, an emeritus professor of psychology at UC Berkeley, has spent many year studying himself, and, even better, offering many practical clues about how to construct your own "experiments of one." I first found out about his work in the most obvious way: searching on "self-experimentation" in Google.

This lead me to Roberts paper: "Self-experimentation as a source of new ideas: Ten examples about sleep, mood, health, and weight." The problems he describes are so common, and his solutions so counter-intuitive, that you can't help being intrigued. One of the great things about reading Roberts is getting a feeling for how different self-experimentation is from other forms of self-knowledge. While Roberts often begins his experiments with a hypothesis, using his stock of common knowledge, suggestions from friends, and categories of analysis typical of a well-trained college professor, this first idea is usually proven, through experiment, to be wrong. Not superficial, or too narrow, or distorted by delusion or prejudice; simply incorrect, provably irrelevant. So then Roberts has to come up with new ideas. The data, expressed as charts, no longer merely test his hypotheses; the data becomes the source of his theories. And the theories bear the mark have having emerged from data. Often, they seem very, very odd. They seem to have no link to received wisdom, to folk knowledge, to intuitive "rightness." To me, they seem like the kinds of theories a computer might have about a person. (A confession: this biases me in their favor.)

Does standing up a lot during the day reduce susceptibility to colds? Go ahead and doubt it; I did. But Roberts has data to back it up, and while it would be foolish to believe that standing up a lot  during the day would eliminate colds across an entire population - foolish, that is, without experiments to prove it - Roberts' own practice of standing up a lot has a lot more empirical back-up than many of the more "sensible" things we naively believe.

Here's anther one: for a long time Roberts had a problem with his sleep. He woke too early, could not go back to sleep, and then was tired in the morning. He tried different ways to cure this problem until, through a combination of coincidence, experiment and analysis of the data, he discovered an expected correlation: his problem disappeared when he skipped breakfast. He cured his early awakening by not eating until 11 a.m.

The idea that skipping breakfast may reduce early awakening was, wrote Roberts, "a new idea in sleep research." Strangely, Roberts was not hungry in the wee hours when he was troubled by early awakening, which lead him to suspect that it was not discomfort that roused him, but rather some glitch in his sleep cycle caused by anticipation of food.
In his paper, Roberts cites a number of studies showing that:

Food-anticipatory activity is a well-established effect in animals (Bolles & Stokes 1965; Boulos & Terman 1980). Mammals, birds, and fish become more active a few hours before feeding time (Boulos & Terman); as far as I know, no effect present in mammals, birds, and fish has ever been absent in humans. Because activity requires wakefulness, food should produce anticipatory wakefulness as well.

Roberts' theory came to mind recently because just last week, in the May 23, 2008 issue of Science, Patrick M. Fuller, Jun Lu, and Clifford B. Saper report on some experiments that precisely locate an important mechanism that links food with circadian rhythms in mice. The idea that circadian rhythms in mice are influenced by food availability is not new, but, through an elegant experiment, the authors show that there is a food-entrainable clock in the dorsomedial nucleus of the hypothalamus (DMH), and that this clock can override the light-sensitive circadian clock in the suprachiasmatic nuclei (SCN).

Our data indicate that there is an inducible clock in the DMH that can override the SCN and drive circadian rhythms when the animal is faced with limited food availability. Thus, under restricted feeding conditions, the DMH clock can assume an executive role in the temporal regulation of behavioral state. For a small mammal, finding food on a daily basis is a critical mission. Even a few days of starvation, a common threat in natural environments, may result in death. Hence, it is adaptive for animals to have a secondary "master clock" that can allow the animal to switch its behavioral pattern rapidly after a period of starvation to maximize the opportunity of finding food sources at the same time on following days.

How strongly this mechanism operates in humans - if at all - is unknown. But, thanks to Seth Roberts' experiments, we have data on a human whose sleep problem was cured by an alteration in the schedule of food. One of the regular contributors to the forum Roberts runs on his Web site uses the tag line: "Proud member of Lab Rats United."  This is a joke, but more than a joke. When we experiment on ourselves, we can fruitfully adapt the methods used by psychologists on mice; but that's not so surprising, because we share a lot of their biology, too.
 
 

Home Monitoring as Long Term Care

Gary Wolf

This story from the New York Times today gave me a sense of how far along the personal monitoring movement has come in the last few years.

...Sensors attached to the wall are able to register when Mrs. Trost gets out of bed and whether she stops at her medication dispenser, and to alert her daughters to any deviations from her routine that might indicate an accident or illness. The family is updated by electronic report every morning.

Monitoring systems like these, which go far beyond the emergency response buttons that have been around for years, are not found in many homes yet. Privacy is an issue for some older people, and the basic package can range from $50 up to $85 a month for the motion sensors and remote monitoring system like Mrs. Trost uses. More comprehensive packages can include devices to track blood pressure, weight or respiration.

Experts on aging say the systems will become commonplace as the 76 million baby boomers approach ages when disabilities or conditions like diabetes and failing eyesight jeopardize the ability to live independently. The population of those 65 years and older is almost 40 million today, and the federal Census Bureau says that will more than double, to nearly 87 million, by midcentury....

The growing number of Alzheimer's sufferers, which is expected to more than triple from the current four million by 2050, may also spur wider adoption of technologies like motion sensors to alert others to deviations in routine, trackers to assure medications are taken and emergency response buttons....

The story by Elizabeth Olson gives a good general description of how personal monitoring is being used to supplement care for the elderly. It only mentions two specific technologies currently in use. They are:

Motion sensors that can inform relatives or care-givers when a person has gotten out of bed, and whether there's been a stop at the medicine cabinet.

Blood pressure monitors and scales that automatically note anomalies and send alerts to doctors.

Examples of the first of these are not hard to find. One of the companies offering them is Simply Home, which sells a programmable motion detecting system designed to identify ominous changes in behavior.

This system is not cheap: $400, plus $54 per month. At first it seems rather Byzantine - alerts go via two way radio to a central processing station, and are automatically send back out to relatives or care givers via the Web and PDAs. But when you realize that there is no call center, no live monitoring, but rather just a data center than processes the alerts and resends them; suddenly, it makes good sense. The data can be interpreted by somebody who knows the person being monitored. Simple rules become useful; if such-and-such a door isn't opened by 9 a.m., send an alert..."

The integration of human watchers into automated systems will be one of the things that makes them smarter more quickly than most people expect. These watchers are not anonymous employees, but relatives and friends, picking up cues from the data stream, and making inferences immediately based on their knowledge of your patterns of life.

Testing Genetic Test Chips

Kevin Kelly

Ann Turner, co-author of the best book on DNA-based genealogy: Trace Your Roots With DNA, wrote me to say that she too has been comparing results from the two big genetic test companies, 23andMe and deCode.  She wrote in response to my earlier posting comparing results between the two vendors.

The big news is that places where errors are showing up are probably not random. Here's the argument, starting with her post on ancestry.com: 

The two companies overlap on 562,532 SNPs. They agreed on 560,128 calls, or 99.6%. 23andMe didn't make a call on 1,970 SNPs where deCODEme did, and deCODEme didn't make a call on 399 records where 23andMe did. That leaves a mere 35 records where they actually made different calls [see the list below]. In all of those cases, one company would make a homozygous call while the other company made a heterozygous call -- there were no cases where they made a completely discordant call.

 
Here's the kicker from Ann's letter to me:

Four of those (rs11149566, rs4458717, rs4660646, and rs 754499) were also found in Antonio's list. That's more than you would expect by chance.

Four out of 23 from Antonio's list and four out of 35 on Turner's list of discordant results indicates that these regions (at least) are unreliable.

This is why sharing results is so valuable and a key to great quantified self understanding.

This is a micrograph of the bead array on which these tests are conducted.

Turner's 35 SNPs with different results, if case you also have done a comparison.

rs10435795 rs1045363 rs10743414 rs10945383 rs11149566 rs11179382 rs11707159 rs11915402 rs1209171 rs1221986 rs12907462 rs1303912 rs13422439 rs161381 rs17328647 rs1961196 rs1966357 rs2016461 rs2064034 rs2290516 rs2853981 rs3952469 rs4336661 rs4423481 rs4458717 rs4572718 rs4660646 rs6531490 rs6942478 rs7102702 rs754499 rs7812884 rs845217 rs9332128 rs9476380

How Accurate Are Personal Genome Tests?

Kevin Kelly

I've had my DNA sequenced by 2 of the 3 companies now offering this service to the paying public. I purchased the tests for 23andMe and Iceland-based deCode. I am still plodding my way through the results -- it's sort of an education. One question I had was how well do the two results matched?  I give the same DNA to both companies; the results ideally should be identical. DeCode claimed to test for 1,000,000 SNPs and 23andMe for 500,000, so the problem of lining all these results up to see what differs is not trivial.  Luckily another user has just done this. 

Antonio Oliveira also used both 23andme and deCode. He writes in his new blog: 

In order to determine the accuracy of the genome profile provided by 23andMe and deCODEme I arranged to be genotyped by both companies and wrote a computer program to compare the results. The downloaded files contains 576,105 snips in the case of 23andMe and 1,013,349 snips for deCODE. After removing the no-calls and matching the two files by SNP identification, 560,299 snips were present in both files. The comparisson revealed 23 cases in which the results do not agree. 

Oliveira made a chart of his results, categorized by chromosome.

The 23 errors makes the agreement between the two sets of data about 99.995% accurate, or an error rate of .005%, which is pretty good for medicine. A better test might be to repeat the test on the same DNA, but I assume the manufactures of the chip have done that. The 23 "unequal" SNPs caught here in disagreement are not SNPs currently associated with any diseases, so these particular errors are inconsequential. I don't know if there are location biases in the errors, but presumably errors can appear in significant locations -- at that very low rate. However if your computer had the same error rate, you'd notice.

"Productivity" Dashboard Monitor

Kevin Kelly

In the annals of self-monitoring tools, here is one that monitors your computer time. It's a fancy version of time management software. You assign certain tags for various functions and websites -- say "surfing" for Digg, Reddit, or Popurls, or "research" for Wikipedia. After you label your activities once, then RescueTime will gather the stats and present you with your accumulative totals in a kind of productivity dashboard. You can get a time budget showing how you actually use time on your computer.

Assuming you can accurately classify which activities are productive, then you can measure your productivity -- at least in terms of how much time you spend on "productive" tasks. (This particular software will also require a certain level of trust since your self-monitoring activities are transmitted to the software's website.) I haven't used it, though it's if free and available on Windows or Mac.

The BodyBugg

Gary Wolf

I'm fascinated by the BodyBugg. Not convinced, but fascinated. This is the most complete self-monitoring system I've yet seen. With an accelerometer, a skin-temperature sensor, a sensor to measure the electrical conductivity of the skin (known as GSR, for Galvanic Skin Response), and a sensor to measure "heat flux" (the rate of heat transfer from the skin), the BodyBugg truly aspires to track a complex behavior – physical exercise – not in terms of outward factors, such as miles run or laps swum, but in terms of inward factors: how much energy has your body used?

This is a hard task, and it's inspiring that somebody has come so far in figuring it out. The goal is round-the-clock self-surveillance:

We recommend wearing the armband as much as possible during waking hours. the more you wear your bodybugg™, the more accurate and effective you will be at maintaining your calorie deficit goal. During low activity level periods of time (such as sleeping), the program will estimate your calorie burn at rest, based on your body parameters, so it is not 100% necessary to wear to sleep.

There is some science available for those who want to calculate energy expenditure through measurements like heat flux. Still, the assumption behind the current version of the BodyBugg is not that users want to experiment on themselves, or participate in scientific research. Instead, they want the Body Bugg to help them lose weight. The problem is that a device that involves such total commitment to rational self-analysis seems ill-suited to such a straightforward goal. If the goal is simply to lose weight, you don't need to measure yourself 24 hours a day, seven days a week. You simply need to eat a little less and exercise a little more than usual. You can track these variables with any calendar program and a scale.

Something like the Body Bugg could clearly do more interesting work. It could show energy expenditure through time, and allow analysis of the relationship between work, sleep, or mood, on the one hand, and patterns of energy use, on the other. It could be used by two or more people, and allow us to test theories of how we influence each other. It could do a lot of fun things. Right now, the Body Bugg is just the technical part of a program of weight loss coaching. But it, or something like it, has a higher destiny.

For people interested in BodyBugg as it is currently intended to be used, there's a good conversation about various issues here. It has apparently been promoted on the TV show, The Biggest Loser, which I've never seen.