As mentioned in my History the Bragi Dash is part of my foray into N-of-1 measurements, using myself as a test subject. Eventually that body of work (and data) will become part of a “health” sub-site under this domain.
The Dash is a compelling device, not only because of its audio quality, but also because of the promise that it holds as an “all conditions” activity monitor. It’s waterproof – and its use as a swim monitor is particularly enticing. Aside from the built-in position/motion sensors, it also had a pulse oximeter – a sensor that could directly measure the oxygen levels in the blood using infrared sensors in the ear canal. While most of the public interest was in its use as a heart rate monitor, I’m interested in the oximeter. So I signed on as a developer for the Dash, so that I could gain early access to its APIs and develop an app that could monitor and analyze blood oximetry data. But I didn’t arrive at this point directly, and as usual – the project has slowed due to external factors.
Bonking into a problem (or two)
One of the things that endurance athletes talk about is “hitting the wall” or “bonking”. This is when the body runs out of glycogen stores (i.e. carbohydrates) during long-term exertion and the body goes into a sudden fatigue cycle. I experienced this when on long bicycle rides, and found it difficult to recover even after consuming “recovery gels”. From that experience I wanted to know more, because those “bonks” were a clear indication that I had to do something to improve my physical training. Through some research online I found the site of Dr. Peter Attia, who described the same scenario in which I found myself. In short, I didn’t just have a exercise problem, I also had a nutrition problem.
For the nutrition side of the equation I am using two measures:
- Capture of food intake through the myfitnesspal app,
- Measure of blood glucose and ketone levels through Abbott’s Precision Xtra monitoring device.
The nutrition tracking provides both macro and micro-nutrient profiles – and what’s more – they have an API that allows me to collect that data from their servers. The Precision Xtra allows direct import of data captured from the device through a special USB connector cable.
To measure my physical conditioning, I’m using both “long arc” and short-term measures, most of which are familiar to people that use activity trackers. I have an all-day monitor in the Polar V800 watch, which is also a highly accurate multi-sport monitor. And during workouts I wear the H7 heart rate monitor, which provides detailed exertion data through my training sessions. Aside from tracking the “usual” activities, such as cycling and gym sessions, it also has the ability to monitor swim arm motion, distance and heart rate. At least, that was the claim. It turns out the the Polar device is susceptible to data drops due to factors such as over-chlorinated water. So there are days when I get really good, consistent data capture – and others with blind spots that are many-minutes wide. So my interest in the Dash was not only for blood oximetry, but also the hope is to have a more consistent means to capture data during all training activities, particularly swimming.
Zone Training and (indirect) RQ measurement
My introduction to respiratory quotient came from this post in Peter Attia’s blog. It had features that I recognized in measured percentage of VO2 max. However, he was going several intriguing steps beyond that. He was factoring in both oxygen and carbon dioxide levels to measure the ratio of glucose and fat metabolism during exercise. The enticing result of this would be the ability to amplify fat metabolism, and greatly extend (and perhaps eliminate) the point at which glucose stores are drained, and therefore prevent “hitting the wall”. As you can see in the graph I’ve borrowed from his site, the fat-to-carb metabolism becomes interesting when the RQ calculation ranges from .70 to 1.00.
While his initial measurement process was decidedly clinical, my current working theory is to take a meaningful departure from that point. As you might gather from the nature of this portfolio entry, I plan to use activity monitors such as the Dash to supply sufficient vectors of information that a reasonable approximation of RQ could be established and measured during workouts. And while that “approximation” method might seem like a dodge, most activity trackers present percentage of VO2 max as a function of an approximation. These devices use age, weight and gender to present a value based on heart rate during exercise. Some smart devices (like the V800) allow the user to baseline their VO2 max directly, using a calibration mode – and so the measure is individualized. But most manufacturers simply use aggregate table data to calculate a result based on an average value. Whether this value is “good enough” depends on the level of training, and the goals of the person that’s measuring their training.
My initial concept is to provide a reasonably accurate RQ measure by taking both heart rate and blood oximetry data from the Dash and using that to calculate a realtime value for RQ. This could be presented as a visual KPI on a mobile (or wrist-worn) device. But more importantly, the Dash can provide voice-based “smart coaching” directly through the earpiece. Those factors led me to this point…
Progress, Promises and Pressing Pause
“So where’s the code?” you might be asking. Well, as these projects go – the Dash was pretty good about delivering on the promise. They were several months behind schedule when the first units started shipping, but they were transparent about their process (both successes and setbacks) and are well into their production process for commercial sales well above and beyond the Kickstarter delivery requirements. The APIs that they promised along with the devices were similarly delayed (in fact, the initial APIs were several months later than the physical device shipments) and still have yet to deliver the API that allows developers to consume blood oximetry data. I would go into further details, but the APIs are provided as part of a developer partnership agreement, which includes a non-disclosure clause.
What I can write about is that the development portion of the project is on hold. There are certain aspects of design that I’m researching, particularly in the area of setting a clinical baseline. This may require going into a labratory environment to take measurements similar to Dr. Attia’s approach, and perform the tests while wearing the Dash – to gather corresponding data which would then calibrate to the clinical data.
The broader hope is that a general model will emerge, similar to the age-gender-weight metric used for VO2 max approximation. If that becomes a reliable metric, then users could choos the “generalized” model and not incur the time and expense to create a baseline in a lab. Until Bragi releases an API that allows access to the oximeter, I will continue researching that track – and will note progress in my blog on this site.