Monthly Archives: July 2015

Personal Gold @ Quantified Self ’15

QS colleague Tim Hanrahan reports from QS15:

Last month’s Quantified Self conference concluded with the premiere of a new documentary on the 2012 U.S. Women’s Cycling Team — Personal Gold: An Underdog Story.

Personal Gold
The Personal GoldQS Panel featuring medal winner Dotsie Bausch, producer Sky Christopherson, and director Tamara Christopherson.

The premise: in the wake of Lance Armstrong’s doping scandal, the women’s team were the only U.S. cycling representatives. They had minimal support for their training leading up to the London games, especially when compared to the multi-million dollar budgets of the U.K. or Australia. So the four women, under the guidance of former Olympic cyclist Sky Christopherson, adopted a ‘Data Not Doping’ mentality to understand each of their individual bodies and personalize their training to cut even the slightest amount of seconds off their time.
It was a fitting close to the 3-day conference because Christopherson led the big data focused efforts by incorporating many Quantified Self studies and applications into tracking the athletes’ response to trainings, sleep habits, and even to the detail of the sleeping room temperature.

Personal Gold

Personal Gold
Athletes were asked to track everything from what they ate, how long they slept, to how they were feeling mentally. All of the data illustrated patterns which were interpreted and adjusted to reach optimal training.

The 90-minute documentary focused on the trials and tribulations of the team’s grueling training in Europe before the thrilling conclusion at the 2012 London games. In addition to highlighting the quantified self processes, we got an inside glimpse at the discipline the athletes needed to compete at the highest level. It was inspiring to see the team’s sacrifice (even the women’s husbands were working full-time to help with the training) and their resourcefulness. Christopherson consulted with numerous big data leaders to learn about the technology and apply it towards the athletes in trial by error fashion. Seeing the errors amidst a heavy time crunch before London only made the result that much more gratifying.

Needless to say, Personal Gold connected with everyone in the QS audience that day. We saw the world’s best athletes use quantified self experiments to improve their optimal self and achieve their goals. I have no doubt that, no matter if you’re a cycling or quantified self enthusiast, the underdog story of the 2012 U.S. Women’s Cycling Team will appeal to you too.

To find out when Personal Gold will be screened near you, visit and follow the film’s Twitter account @Personal_Gold. Without further ado, watch the trailer below to drive home everything above.

Sign up for the QuantXLaFont Newsletter
Get our lifestyle tips and studies delivered to your inbox.
Thank you! We don't spam :)

Exercise and the Daily Energy Curve

One critical task that I solve for is ensuring that I maximize my energy during daily negotiations. I want to be present and balanced when I engage with others and not have my physiology defaulting me to Vapor Lock because the night before I had a bad batch of cannoli. So I look to maximize my Daily Energy Curve.

To measure my state of balance I use Heart Rate Variability (HRV).  I know that my balance and energy goes down over the course of a day as I shared in a past post. If that was the case, was a hard workout in the morning depleting my energy before I went into negotiations?


Logic said that using energy early in the day would leave less for the remainder of the day. So I had to test it.

My Question

Was exercising early in the day lowering the remaining energy I had for the remainder of the day when I would be in negotiations with others?

The Resulting Potential Action

If exercise first thing in the morning had no effect on my energy levels I would continue to exercise in the morning. If it did have an impact on energy levels on the days when I was engaged in important negotiations I would either skip the workout or workout later in the day after the negotiations.

What I Did

I created a random list of Workout/No Workout days to ensure that the results were not skewed by some personal bias. On days I was to workout, I exercised for 30 minutes on an elliptical machine either at my house or in a hotel on the road.

I took a measure of my HRV at least three times a day, one on waking, one in the afternoon and one on going to bed in the evening. The combination of these three measures I put into a Google spreadsheet and calculated the slope of the three measures. To measure my HRV I used a Polar H7 heart rate belt, an iPhone6 and the Heart Rate Variability app.

I took 25 readings over the course of a month. 19 of the readings were from randomly generated instruction, 6 were due to life events (elliptical not available, had an opportunity to workout).

What I Learned

The difference in my HRV slope on days I exercised had no statistically significant difference than on days that I did not exercise. There was no correlation between exercise and my energy levels.

The idea of the HRV slope reflecting my Daily Energy Curve which would steadily drop over time  means that we would expect the slope to be nearly always negative. Supposedly HRV starts high and ends low. With that assumption, the workout would drop the early day energy and the negativity of the slope would increase. The Daily Energy Curve in the Exercise or Not Exercise would look something like this:

Slide1In the actual readings, there were 9 days that the slope of the curve was positive, with close to an equal number of those positive days being on both Exercise and No Exercise days. You can see the positive days in the scattergram of the readings over time:


And a scattergram of the Exercise (1) vs. No Exercise (0) showing the distribution during the study:

Slope Exercise or Not

Running a few statistical tests on the data it came back that there was no difference between the HRV slope on days of exercise and a random sample of HRV slope readings. On both a T Test and Pearson Correlation the difference was not significant.

So in a (semi) randomized test of exercise effect on the Daily Energy Curve I dispelled a cherished personal myth. In the past, when I would wake up on the day of an important negotiation I would say “I should save my energy for the discussion” and blow off the workout. A study and a bit of math now tell me that I’m not really saving up energy if I skip the workout. I’m just blowing off the workout.

I have to give full credit for the randomization method to Cara Mae Cirignano of Whatify. I did use Whatify for a portion of this study but because I did not get the entire study done with them they are off the hook for methodological irregularities. I highly recommend you check out their service.

So going forward my workout decisions are independent of my pending negotiations. And I have to take more of a look at the HRV Slope. If sometimes it is positive, what is driving that? For a future study.



Measuring Muse Mindfulness vs HRV

The more experienced you get with taking your own measurements you will find that even the smallest technique change can alter readings. A wary Quantified Selfer will be aware of this and control for variation. I found this when I measured Muse Mindfulness vs HRV.

Early in my self-measuring career, I discovered that real-time data can cause a stress reaction I called Freakback and I even tried to describe its  anatomy. When your reading is telling you that you are not relaxed you try to force relaxation and that causes stress which raises the stress reading. Eventually the amplitude of stress reaches “not inconsiderable discomfort.”

Each morning I measure my heart rate, EEG, glucose and ketone level. From these readings I get the measure of how much energy I have for the day ahead and can see the impact of the previous day’s activity and sleep on my physiology.

Muse Mindfulness vs HRV

My Question

When taking my EEG and heart rate, is it better to take the readings simultaneously or have one precede the other?

The Resulting Potential Action

Based on the result I would alter my morning meditation session to either take the heart rate and EEG reading together or  take one reading before the other. This resultant technique I would consistently use to ensure any variability came from external factors rather than my changes in data gathering technique.

What I Did

To measure my heart rate I used a Polar H7 heart rate belt, an iPhone6 and the Heart Rate Variability app. The output was my resting heart rate, rMSSD, Low and High Frequency of Heart Rate Variability (HRV).

For my EEG, I used the Muse headset connected to an iPod5. This gave me a %Calm score after taking a baseline of active thinking.

First, I took readings simultaneously for 15 mornings. This meant I calibrated the Muse, turned on the HRV reading then breathed rhythmically using a breath pacer. At the end of 5 minutes I had both readings.

Next, I took the Muse readings first, then HRV for 10 mornings. The difference here is that during the Muse readings I was counting breaths with eyes closed, during HRV I using a breath pacer. So my method of concentration was different unlike the simultaneous readings.

Finally, for 13 mornings I switched the order with the HRV session first and the Muse session following.

After each session I entered the values manually into a Google spreadsheet for analysis.

What I Learned

The most efficient way for me to measure my EEG and heart rate in the morning is to conduct the readings simultaneously.

Because I could only look at correlations rather than causes in this study it was important to me that when I was calm and balanced the two readings would show some consistency. For example, if I had a poor nights sleep I would think that my %Calm and rMSSD readings would both be lower. So the correlation of the two readings was important to me.

Taking my HRV and %Calm reading at the same time had a strong positive relationship as you can see here from the Pearson correlations:

  • HRV before Muse; r = -.14 (No relationship)
  • Muse before HRV; r = -.77 (Very strong negative relationship)
  • HRV same time as Muse; r = +.48 (Strong positive relationship)

The other methods show interesting differences in readings based on a variance in technique. It seems logical that an upset physiology would show low readings in the same 10 minute period. Or vice versa.

Instead when I shifted from simultaneous readings to using Muse first, there was a very strong negative correlation. That meant when I had a session with a high %Calm EEG reading a lower HRV would immediately follow. Or when I was thoughtful and had low %Calm my HRV was higher. I am open to comments on why that may be. It is a unique outcome that needs further testing as there may some interesting things going on.

When I took the HRV reading before the Muse reading there was no relationship. Somehow that sequence of reading scrambled the signal because there does seem to be a relationship between %Calm and rMSSD when read at the same time. So I have discarded that technique.

As a quick check I compared the average %Calm reading with an average scaled rMSSD (translating it to a scale between 1 – 100) for each of the three techniques. Here is the result:

Muse Mindfulness vs HRV

The average rMSSD was close across the board (45 – 51%) but the %Calm was much lower for the HRV before Calm and much higher in Calm before HRV. Some ideas for this may be that the time pressure of wanting to complete the reading and get on with my day may impact the mindfulness. But that is for further verification.

So with these results I will use Muse and my HRV apparatus simultaneously to measure the waking state of my physiology. I will set up the process so it is automatic and consistent, and begin testing external influences on the morning state.

Booze, HRV and Muse

I’ve become interested in finding with greater precision how my body reacts in different circumstances. I had a story that I felt poorly after eating cheese. It was based on one period of my life and was entirely anecdotal. As I described in my post about my shift to a ketogenic diet, I had started eating more fats, to include cheese. And I felt great. This got me thinking about how many of these unexamined stories guide my behavior.

With that in mind I started looking at alcohol’s effect on waking mental calm and heart rate variability (HRV). I like a glass of wine or two in the evening. It is well documented that alcohol has a physiological impact. From quantified selfers looking at their Basis data to scientific studies there is a wealth of information on the physiological impact of alcohol. My question is how much can my body tolerate before it reduces my heart rate variability and mental calm.

Booze, HRV and Muse

My Question

What impact does having alcohol (or not) have on my morning HRV and mental calm?

What I Did

For thirty-three days on waking I measured my mental calm and heart rate variability while noting if I had had alcohol the night prior.

How I Did It

I would measure my mental calm using Muse EEG headset and my HRV using a Polar H7 heart rate belt that sent data to an app called the Heart Rate Variability Logger. All data went into a Google spreadsheet.

What I Learned

Drinking alcohol the night prior reduced average HRV and increased average Muse “% calm” score. After a night of boozing I was mentally more Zen but my nervous system was under increased load.

The Muse % calm readings were not visibly different between having had alcohol or not. The Pearson correlation (.12%) showed a negligible relationship between the calm readings and whether I had consumed alcohol the night prior.


Booze, HRV and Muse Booze, HRV and Muse


My average % calm after having no alcohol was 36.4% and after having alcohol was 42.7%. Looking at the averages I appeared more calm after drinking. Because the scores did not correlate with consumption we can’t draw any real connection.

My HRV did correlate with consumption, with a Pearson correlation = -.306, a moderate negative relationship. There seems to be a connection with drinking alcohol and a lowered HRV was lower the following morning.

Booze, HRV and MuseBooze, HRV and Muse


My average rMSSD the morning after a night of alcohol was 48.6. A morning after no alcohol the average was 62.1. A moderate correlation and a much lower average verified that HRV seemed to be impacted by alcohol.

But knowing that does not really give me any precision nor any guidance on how to change behavior. So my next study is to examine how much alcohol I can consume and not have my HRV drop. If I am going to enjoy a glass and don’t want to take an HRV hit the next day, how many glasses can I have?  With this I will know my own tolerance and be able to guide my actions with better data.

Going Ketogenic and HRV

Inspired by a fellow Quantified Selfer at our recent Denver QS Meetup and motivated by multiple guests on Damien Blenkinsopp’s excellent Quantified Body podcast I decided to switch from eating vegetarian to eating ketogenic. And always interested in heart rate variability (HRV), I wanted to look at the mix of going ketogenic and HRV.

I won’t go into too much detail about ketosis and its reported benefits. At a high level you switch your body over from sugar (glucose) burning to fat burning (ketones). To do this I was going from a very carb heavy diet to one that severely limits carbs and moderates proteins while ramping up fat. Imagine my long suffering vegetarian wife’s surprise when she came into the kitchen to find me chomping on bacon with a coffee full of butter.

One element of the switch involves a state called “keto flu” which is the body suddenly finding no carbs available and sending pretty clear signals to the brain that its time to find some carbs. Proponents of the diet tend to downplay the effects calling it a euphemistic “short term discomfort” but I have to say it felt like I had been hit by a truck.

Going Ketogenic and HRV

Me no like keto flu

Of course being a Heart Rate Variability (HRV) enthusiast I had to check the impact of the transition on my resting heart rate and HRV.

My Question

What impact would the transition to ketosis and its associated “keto flu” have on my HRV? What did the load on the body look like?

What I Did

I measured my resting heart rate and HRV for three weeks during the transition to ketosis. I was also measuring my glucose and ketone production so could verify when the transition was complete.

How I Did It

I measured my heart rate and HRV reading using a Polar H7 heart rate belt that was sending data to Marco Altini’s Heart Rate Variability Logger four times a day. I would take a reading first thing in the morning, later in the morning after traveling to work, after lunch and before bed. Once each morning I would use a glucose meter and a Ketonix breath ketone analyzer to check the trend in my transition to ketosis. Toward the end of the period I used a Precision Xtra ketone meter to verify with surety I was in ketosis.

What I Learned

Transitioning to ketosis is a real butt kicker of both mood and heart level activity. First my mood went very dark and on day two I almost tore my car’s steering wheel out of its moorings when I couldn’t find a parking space. Keto flu for me was like a bad hangover mixed with a flu-like dizziness. For me it took two weeks to transition to ketosis.

At the physiological level my resting heart rate over the period showed a large amount of load on the body. You can see from the graph that my resting heart rate popped up almost immediately. In the middle of the transition I was getting a readings of 85 to 93 Beats Per Minute (BPM) and that was unusually high. I am usually in the high 60’s to low 70’s.

Going Ketogenic and HRV

HRV showed a similar story. The four readings through the day showed that my HRV was lowered in the early day readings but tended to be similar in the later day readings.

Going Ketogenic and HRV

HRV has a circadian rhythm. As with our energy levels, it tends to drop over the course of the day. You can see that in the baseline. The slope of the curve was still negative while in the transition, but the slope was greatly reduced. I would have expected the slope to be the same and the lower start point end with a lower end point, but that was not the case. Perhaps the body has a baseline beneath which it won’t drop based on health and when in load the impact is in the beginning of the day when the body should be able to relax.

From an HRV perspective it looks like transitioning to ketosis has a similar effect on the body as being overtrained as an athlete.  The body’s Parasympathetic Nervous System seems to be less capable of putting the brakes on the system and bring the body to a relaxed state.

The happy news is that once through the whupping you get from keto flu the benefits are as advertised. I dropped 12 pounds and feel great. I have a lot of energy and feel very clear of mind. All the better to conduct more studies in the future.