Landmines & Zombies
Project Infobox  

Self researcher(s)  Chris Bartley 
Related tools  Excel, chart 
Related topics  Food tracking, Stress, Mood and emotion, Medication habits 
Builds on project(s) 

Has inspired  Projects (0) 
Show and Tell Talk Infobox  

Featured image  
Date  2013/10/11 
Event name  2013 QS Global Conference 
Slides  Landmineszombies.pdf 
This content was automatically imported. See here how to improve it if any information is missing or out outdated. 
Landmines & Zombies is a Show & Tell talk by Chris Bartley that has been imported from the Quantified Self Show & Tell library.The talk was given on 2013/10/11 and is about Food tracking, Stress, Mood and emotion, and Medication habits.
DescriptionEdit
A description of this project as introduced by Quantified Self follows:
Chris Bartley, an engineering consultant was feeling like a zombie. So over the course of 18 months, Chris tried taking supplements and medications and made changes in his diet. He started a spreadsheet with five columns to track his wellness, medications and diet. He saw no improvement and decided to write a thesis on himself. He took the data and decomposed it into indicated variables. He found that it was only a subjective measurement and learned that he does know his own body.
Video and transcriptEdit
Chris Bartley  Landmines & Zombies
Hello, my name’s Chris and it’s just amazing to be here. That talk very similar to my story except I’ve got zombies in mine so take that. Look it’s just great to be here. I’m from Perth in Western Australia, unless you think it’s a back water it’s actually more expensive to live in Perth than in San Francisco, a cup of cappuccino cost’s about 4:50 in Perth and 3.70 here or so I’m told. So my story starts at the University of Western Australia where I did Bachelor of Engineering and Science, and my thesis was on landmine removal in Afghanistan and got to go to Pakistan for three weeks as part of that in 2001. So landmine removal is basically done with a metal detector, and whenever the metal detector goes off they get down on their hands and knees to find out what it was which is usually a piece of metal, a shell, bottle cap. Unfortunately for me, while I was there it triggered Reiter’s syndrome or reactive arthritis which seems to be autoimmune. And I got over that and I could walk again in about three months, but after that I was tired all the time. To try and explain what it was like, it felt like before my mind was like a Ferrari, but afterwards it was a struggle to push it up to five miles per hour. So I finished my degree. I did well and I won the engineering prize. I started working as an engineering consultant, but little to what most people know is I looked normal, most of the time I felt like a zombie. So I went to the doctor like many others here but everything was fine, all good. I tried the internet. Very similar didn’t work and then moved on to try anything and then I just settled into a world of fog. Amazingly, a wonderful woman married me, Liz and they made a movie about it recently, and we had our first boy six years ago and that really triggered me to kind of get back on the coach and find out what was causing the brain fog. So I went to a doctor who specialized in chronic fatigue, and he did some more specialized tests including micro bone, hair, blood. And basically over the course of the next 18 month I tried about 30 different supplements and medications and a few changes in diet. And I also started a simple spreadsheet with five columns to track my wellness, my medications and diet. My wellness is simply a subjective measure of how I felt out of 100% in terms of brain fog. So after about a year and a half there was no clear improvement and it occurred to me that I came to the end of my ideas, and why don’t I use my thesis on myself. So this model was actually developed for landmine removal and it’s the time, the man hours it would take in terms of the main factors effecting landmine removal, which is mainly the number of fragments and land use type and a few other things. So anyway, I took my data and decomposed it into indicated variables. So there’s 38 variables, one output being my wellness, an indicator if it was a holiday or not, five dietary indicators and 32 supplements. And then I applied multiple regressions to it. so a stepwise procedure to either optimize the variable form or eliminate it if it wasn’t statistically significant. So of the 38 inputs only three were statistically significant. And the first was holidays, and you can see the red indicator for holidays down the bottom. So mostly just weekends but there are three stretches of holiday there, and if you look up the top the p value is very very low so the statistical significance of the correlation is very high. The coefficient, which is the effect, is about 7%. On average it lifted my wellness of about 7% when it was a holiday and the optimal form was on the same day of the holiday so not the day before, not the accumulative effect. And this was a really encouraging result for me because it gave me confidence that the procedure was working. I think this is basically a property of stress which I wasn’t tracking directly. This is a simple boxplot showing comparisons of holiday and nonholiday. The red line in the middle is the median and then the box extent is the first and third quartile and then the outliers so you can see a clear difference. Not the second result was one of the 30 odd supplements and medications that I had taken. Only one was statistically significant and that was tyrosine; an amino acid supplement. Again you can see up the top the p value is very low; a high statistical correlation. The coefficient is about 7% again, so average of 7% lift in wellness and the optimal form is on the ame day I took it again. This was a very exciting result for me. you can actually see and having been pointed out by the algorithm, you can see how the blue is bottoming out, and perhaps you can see it better on the boxplot; the lift in the median and the dropping in the range when I was having tyrosine. A very clear result and it was just exciting that the algorithm, the multiple regression picked that out of the noise. And the final one is red meat, and red meat is bad. I don’t know why, and the p value is slightly higher, it’s 7% so the certainty is lower. The effect is negative; 6% and the optimal form is actually accumulative on the same day, so it seems to have an accumulative effect. So to summarize, the holidays are good on the same day, tyrosine is good on the same day. Red mea tis bad, and the accumulative effect with a lower level of certainty. So this was about two years ago and I got pretty excited because of the way it just lifted results out of the noise was really exciting and it provided some concrete ways to go forward, and I’ve actually designed an app just for my personal use and I’ve been tracking food and I’ve just broken it down into all of its components. And I’ve got about 10 months of data and I’m in the process of doing a slightly more advanced statistical analysis. But I didn’t finish that in time for the conference so hopefully I’ll present that later. So what did I learn? So I think the key question is, why did it take me so long – eight years to realized that I could apply my own statistical skills to my own life and my own problems. I think that I thought that the medical profession was my only hope. But I learned that there are things I could do myself. I think I thought chronic fatigue was multicausal and too complicated, and that’s exactly what statistics are for. I think I thought that the data was too rough, but with statistics, rough and regular is enough. And I thought it was only subjective measurement but I learned that I do know my own body.
So that’s it, thank you.
About the presenterEdit
Chris Bartley gave this talk.