I started tracking myself after being warned by my primary care doctor about the complications of my obesity, which is elevated blood pressure and glucose level. He recommended that I should exercise or at least walk, which I am at least certain of. He also probably told me to eat less, although I can't quite remember. These pieces of advice are unhelpful, as they are not actionable. Many people lose weight only to gain it back once temptation takes over.
A much more substantial prospect in my mind is the idea of self quantification, which is the art and science of conducting scientific experiments and data logging on one self, for the purpose of self improvement. This was an idea that had been lingered passively in my mind until recently when The Atlantic ran a story about a computer scientist who self quantified himself and discovered that he have Crohn's disease before his doctors. This story hit home with me.
A few years ago, I had undiagnosed Cohn's disease that went on for a few months. The doctors made several recommendations to no avail. The side effects of the medication just made it worse. The following summer, I was rushed to the ER after I suffered internal bleeding. The digestive tract had developed an ulcer which had opened a blood vessel(I am not sure if this is the correct mechanic). Eventually, a new GI doctor did an endoscopy on me, which gave me a diagnosis of crohn's disease. Then the ulcer came back the following school year and led to two more hospitalizations. I have been very fortunate to not experienced another hospitalization in the two years after the last incident, however I constantly worry that it will happen again.
If I had been monitoring my health continuously using all sort of instruments and lab tests that the computer scientist did, I may be able to detect the inflammation in my body and be able to treat my Crohn's disease early, allowing me to avoid one of the more unpleasant complication of the disease. However, due to the limitation of the technology and limited access to these technologies, I probably lack the ability to access many of the data that the scientist have access to. In this sense, even if I begun the self quantification since the beginning of high school, I don't necessary have the capability to detect the disease as early as the scientist. Moreover, the tech may be measuring what I need to know, but it could be expensive, unavailable, hard to use, inconvenient, or heavy. That impedes people from self quantifying as much as they need. The computer scientist went to great length to get tests to get all sort of data, but it doesn't mean he necessarily like all the procedures. For example, to do a proper endoscopy, you have to drink lot of nasty tasting fluid to flush out the content of the digestive tract. To get an MRI scan, you have to at the very least drive to the MRI center.
Even with all the information gathered, it doesn't mean you will understand the processes in your body. One doctor who was mentioned in the Atlantic article fears that self quantification will lead people to over-diagnosing and over-testing because of unexplained anomalies in the body. True that there will be unexplained anomalies, but these anomalies just mean that our medical knowledge is inadequate. You only visit the doctor every once in a while, the rest of the time you are unmeasured and unobserved. Without data, there can be no knowledge.
Despite the limitation of technology and knowledge, it doesn't stop self quantification from being a useful part of my life. At present, I measure my blood pressure, heart rate, weight, and how many steps I walk. Being able to know those basic information help keep of my health condition. Another benefit is that if I have a goal, I will often strives to reach it. Walking 10K steps a day is my first goal in my self-quantification program, and I manage to keep a goal reaching streak for as long as 37 days. For example, I often have high pulse rate, which let me know that I need to eventually start a cardiovascular exercise program. In the future, I could also measure my mood on a notecard, track what food I am eating using my phone, how many pushups I did, and so on. All of these yield information that I can use to guide my decision-making to improve myself. More automated and expensive tools will improves my tracking ability, certainly, but sometime the cheaper or free tools are all what you can do.
Still, there's one thing I can credit to my doctor: he suggested the idea of using a pedometer. I already have that pedometer in mind by that time too. The doctor's recommendation only gave me a kick in the direction I needed to go. A few days later, I began the effort of exercising and quantifying my steps. Somehow, the self-quantification and the exercise effort is sticking far longer than any of my previous attempts ever did. I am still going strong, still counting my steps since I started way back on August 26, 2012. I already finished one experiment and will be working on more.
The rest of the essay is about the details of my self-quantification effort such as goals I have, tools I use, how I gather my data, and finally, a series of interventions. Each Intervention section have a basic layout, divided into three parts. The first part describes what I am hoping to accomplish, my hypothesis or intervention, and my confidence level about success. In the middle, I'll gather up all the data and analyze it. Lastly, I will reach a conclusion regarding if it is inconclusive, negative, or successful, as well analyze the shortcoming of the experiment if any. It's not a guide on how to run an experiment and how to quantify one self over time. It may be something you might want to use as a guide for doing something similar to the experiment.
Probably the greatest single most important factor is your health. Your health decides how much energy you have, how long you live, and your state of mind. Therefore, your health impacts everything else. My health is below average, but not very poor, even though I neglect it. Since it's the most neglected of any part of my life, that what I want to focus on first.
The first health issue is my weight, since obesity correlates to everything else that is bad such as sleep deprivation, diabetes, heart diseases, high blood pressure, and cancer. Once I started measuring myself, I learned just how true it is. Before I started tracking, I weighed 206.6 pounds and was 5 feet 7 inches tall. My BMI is calculated to be 32.4, which is considered obese. My long term goal is to reach the BMI of 24.9, or 159 pounds, which is considered normal weight. 1 No
Once I get my weight down, I will be able to focus on other aspect of my health, such as my cardiovascular performance, or sleep hygiene if I felt I am sleeping poorly. However, once good health is achieved, it's very likely I will switch focus to other area of my life such as my finance or my work productivity. For now, I am just focusing on losing weight.
The way I choose tools is based primary on cost and accessibility. A pedometer is 5 dollars. Electronic scale and a blood pressure cuff are what I have in my home. The tradeoff of cheap tools is that they lack functionality that aid in self quantification, but they are more accessible to the average user. My incredibly cheap pedometer cannot do anything other than have a goal and counting. It doesn't reset on midnight, nor does it keep a 7 days log, or have any other such convenience. Fitbit, one of the most expensive model on the market, appears to be the most convenient. It will automatically sync to your laptop and smartphone, record sleep data, and count how many flight of stairs you walk. However, I can't afford them on my budget.
Of all the measurement that you can take, counting steps is probably the easiest and one of the cheapest. You can simply wear the pedometer and walk around. Step counts are available immediately, which allows you to make decisions throughout the day when to walk more. The hardest measurement so far is taking your blood pressure and heart rate. You want to be resting, and you need to make sure the blood pressure cuff is tight. However, it not only takes your blood pressure but also your heart rate. So you get two piece of data at once.
It is my policy to add new tools every 21 days in order to form habits with the new tools. If one tries to adopt new tools too fast, cognitive burden will increase, leading to an increased likelihood of dropped tool use, and consequently, dropped data collection. With longer timeframe between each new tools, tool use begin to become automatic, leading to new habits and decreased cognitive burden. 21 days is probably not long enough for any habit to become automatic, and the days that it take to form a new habits will vary on the difficulty of the task. Still, 21 days is what I judged as "good enough". 2 However, there's also instance in which I decide not to adopt any new tool, but rather just new data. For example, waking up and sleeping time does not need a fancy tool, although it does help to have something like the Zeo.
No matter how automatic my habits are, each still represents a cognitive burden. As I add habits to my routine, my cognitive burden will increase. At this point, I only used 4 tools to measure, so the burden is low. Tools that are convenient or requires less effort to form habits will become an increasingly important quality as I add new tools or decide on tool replacement.
Currently, I am measuring steps, weight, blood pressure, and heart rate, glucose level, and sleeping time(including naps). Each additional data points are originated from a particular tool that I am using. Naturally, the more tools I am using, the more data points there are.
For the first week, I inputted my data on two separate paper covering two different set of data, steps and weight. In hindsight, I should have input all my data in my laptop instead. At the end of the week, I input the data initially using Gnumeric and saved as a csv file instead of the default xml format. However, Gnumeric doesn't really like saving in CSV format, so I switched to Libreoffice, which I have been using for the rest of my data collection effort. Then I made a new repository on github so that any tools and data I built will not be gone if my laptop dies.
Initially, I wrote a command line application to access the data and analyze it, but I never really went any further with it. Eventually, I thought of exporting all my spreadsheet data into my website, where I can build custom made forms for entering data and automating some data input task. I am still working on proof of concept export and import code to do this, but I have some of the system ready for entering steps data and the like.
Sometime, there are gaps in the data, usually because I forgot to take measurement for that day. Sometime, I would not know where my tool is, so I lost data. To fix this problem, it's better to have automatic data collection and be something that I wear, preferably with the ability to stick to my skin. If I change my clothes, I have to remember to put back my tools on. For example, it's better to have a pedometer that can sync to your laptop rather than having a pedometer that you must read and input by hand.
There are also other data points but they are dispersed. These information includes editing metrics on this site, learning data on KhanAcademy, anki stats, and much more. I hope to someday consolidate them into one master spreadsheet.
My first intervention is to walk 10K steps a day. While I am hopeful that I will lose weight by walking everyday, I am skeptical that intervention alone will decrease my weight. There's also the issue that any number of factors would influence my weight, such as change in physical activity level other than walking, or change in diet. Nonetheless, I believe I can change my health by increasing my activity level from sedentary to light. 3 I figure that I need to complete at least 30 days to have good data. I started on August 28, 2012 finished gathering data on September 26, 2012.
The data consists of weight and the time and steps in a 24 hours period. Steps don't have time logged, due to oversight. Since I only took off my pedometer at night after I stopped walking, we will assume all steps are logged at 11:59 PM. Admittedly this is not perfect, since I could have taken it off at 2 AM in the morning, but I more or less usually achieve 10K steps by midnight. Sometime, I forgot my pedometer when I goes out, so the step counts may also reflect undercounting. Since the experiment was conducted early in my self quantification history, blood pressure data and heart rate was only logged in the last week of the 30 days experiment. However, since there were no new intervention, we were able to get more than a month of data.
Data analysis and presentation of data are not available at this time, because I choose to wait until I wrote the tool to analyze and output data, rather than do everything manually. The kibabase website will need to add support for graphs and charts, while the tools available for analysis and output will need to be developed as well.
Despite the lack of full analysis, I am lead to conclude that walking 10K steps have almost no effect on my weight loss or a very subtle one.
To achieve my goal of walking 10K steps, I am resorting to creative ways of achieving high step counts. For example, I park my car far away from the entrance of a store, which forced me to walk more. When my mom forgot her lunch once, I volunteered to accompany her to the shop even though she doesn't need me. When I am waiting for my mother to get off work, I usually take the opportunity to walk around the whole store. When a store is not yet opened, I also walk.
I used to walk outside, but I cannot do anything but think idle thought while walking. So, I began walking inside my home. Since the home is a less a dynamic environment than the neighborhood sidewalk, this allows me to focus on other things. At first, I used my phone to chitchat, but I found that to be boring. Then I started reading books on my laptop while I walk around the house. Carrying the laptop around is a pain due to the weight, but I am able to reap more benefit by increasing my knowledge of the world. This reading habit help grows a page of book reviews. Often time, I was able to read at least a hundred page a day. Sometime I finished a book in one day because they are so interesting.
I read two books in nutrition by Gray Taubes, who suggests that carbohydrate is uniquely fattening. In his books, he wrote several reasons why and point to convincing evidence as to why carbohydrate are leading to metabolic syndrome and why mainstream nutrition science are wrong. I am highly convinced by his writing, but it goes to say that his opinion is not what the scientists in public health are proposing. I am not making the claim that we should believes in authority blindly over Taubes, but I am assuming scientists at public health organization are data driven and posses the proper framework for scientific inquiry. Anyway, the best way to settle this question is to actually conduct an experiment.
It is impossible to run this experiment without randomized trials simply because I am one person. It's also impossible to do a placebo because taste will differ. These are once again, the limit of self quantification. The sample is exactly only N=1.
In this experiment, all food that is destined to become dinner will be documented, large and small, with a smartphone camera. These data will be then uploaded and categorized for future analysis. Unlike the previous experiment, I now have ample amount of data before the experiment to compared. However, I do not have any data pre-diet regarding my food. I can only tell you that it involves rice as staple food with no attempt to control carbohydrates.
The variable in this experiment will be a prohibition of all bread, potatoes, and other starches, except rice. Anything that is either green vegetable or meat can be eaten freely. However, I do not have any control on food spending, so consistency and adherence to diet will be hard to achieve. However, the amount of rice will be subjected to consistency. In the first weeks, 3 bowls of rice per day is allowed, as measured by a small bowl. After each week, the amount of rice will reduced by one bowl until there is no more rice in my diet.
The duration of this experiment will be 4 weeks + extra days that I did not conduct another intervention. Preferably, my diet will not change really fast to allows for really good data gathering, especially if all the noise in the diet are removed. I have no specific start date in mind, other than it will start as soon as the initial analysis of the first experiment is complete. If we're lucky, we may even see pre-experimental data on the typical diet I have before the start of the experiment, since I always add more data points every 21 days.
An article on the variation of length in habit formation.. Accessed on September 29, 2012↩
A 2007 Meta-analysis suggests that walkers will decrease blood pressure independent of BMI. Hence, even if I don't lose weight, I may still get healthy by merely walking a lot. Sadly I do not know my activity level from before the self-quantification and intervention, but I do believe that I changed my activity level significantly. However, the meta analysis is rather contentious, saying pedometer based studies are not consistent in their experimental protocol, do not have large enough sample size, and the studies were too short in length to confirm health effects in the long term. Last accessed, September 9, 2012.↩