The NIX Hydration Biosensor for Triathletes
Note: you can skip to “The Review” if you don’t want background blather. If you don’t care about a functional review and just want to know if this thing works (sorta), go to “The Data”
I have been triathlon training for about 3.5 years. I’m no expert, and I really don’t have any hot tips as I am still listening, reading, learning and sharing. I feel pretty confident in getting someone started, but not how to compete as an elite. I am no elite triathlon athlete.
However, as an engineer, I do enjoy the gadgets! Not just the gadgets themselves, but the way I use them to measure and track performance. Training Peaks is my epicenter for data storage and analysis, even though Apple does a great job of storing information and providing me health related trends that I do not get from Training Peaks (like VO2 Max, and other things).
My most recent epiphany is that I am not doing nutrition right. After my first 70.3 race, when I started the run I told my wife, “I got nuthin…no energy”. I walked most of the run and then lay down at the end and fully cramped up and could not get up for half an hour. That never happened again, but I always crap out on the run. Great swim, good ride, bad run. Every time.
That is basically me (sans what looks like milk coming off that guys chest). People sweat at different rates, and with different concentrations of salt. I have not yet done a professional trial measuring those two metrics, but I know the salt content of my sweat is very high. My kit after a race looks like I rolled around in salt trying to turn myself into gravlax. I now suspect I need near 3 grams of salt per hour during a race. That is extremely high.
Getting this measured professionally is expensive and inconvenient. So I was pretty happy when I found the Nix Hydration Biosensors. These little devices stick to your arm, and measure both salt content and sweat production rate while you are training.
I looked for reviews on Youtube and on text platforms, and ALL of them were vapid reviews that I can summarize as “I turned it on, and it said a number”. DCRainmaker, probably the most reliable place to get product reviews, didn’t cover the Nix. So, I decided to do one. I am also including suggestions to Nix in case they ever read this.
The Review
I will start by saying that the concept is ideal, for those of us who want to understand this. Let’s go through each part of it.
The Hardware
The Nix comes in two parts. There is a black clamshell box that you put the Nix ‘pod’ into to charge it up. It’s pretty straightforward and I never had any issues charging the device. But I did come to my first frustration.
The pod is axisymmetric on the top. The case is also axisymmetric. There is no labelling, the company name is seen in the molding. All the parts I need to fit into each other are black.
You can’t see it in the picture, but the charging case is about 6 times larger than the pod. So you charge the pod, take the pod out of the case, stick it into a patch, and stick the patch to your arm. Simple!
Except all of those steps are annoying because the devices are axisymmetric, black, and hard to see the orientation. Even just opening the case is annoying because the hinge is totally hidden, it takes many cycles until you learn where the USB connector comes out relative to the hinge, because you can’t see a label. The pod is also axisymmetric, black and hard to see the correct orientation, while the keyways below are NOT visible at all. I found myself flipping it over, looking for the big molded key, then finding the big keyway on the patch to be able to attach the two. This is true for putting the pod back in the case. The design looks neat, but it is very cumbersome and annoying to use.
Suggestion #1: Lose the case altogether. Put the USB-c port in the pod itself with a small door (like the Garmin Edge devices have) or a silicone plug. Maybe just include a protective cover people can use to put over the side with the connection pins. Cheaper! Way less annoying.
Suggestion #2: Use a colored silk screen to indicate “This way up” on the pod, the case AND the patch. Even better, lose the axisymmetry and have the shape itself indicate orientation.
The Software
The software is pretty standard. Make an account, sign into the app, connect the pod, start your exercise and go!
The first screen above shows how much sweat you have lost and should replenish. The second screen is an acitivity summary with regard to water loss and electrolyte loss. And the third screen is a list of activities so you can go back and check stuff out. These are pretty straight forward and what you would expect.
Nix also has something called an NIX index (oh yeah, totally trademarked). I get it, you want to capture some IP space.
The idea behind the Nix index is that they have combined a bunch of weather metrics (temperature, dew point, things like that) to create a number. So if you know upcoming weather conditions, you can go back and find those weather conditions and see how your water and eletrolyte losses looked like. Cool, right? What? No.
What’s the index for? For this to even be useful, I need a map where I can pick a point, then the app should tell me the predicted NIX index for that spot on the specific day I need, race day. After it tells me the NIX index for that, then it should look through my activities for a similar NIX index and recommend a water and electrolyte strategy. It doesn’t do any of that. I basically ignored the NIX index altogether. Further, I, like many triathletes, train indoors often for biking (no NIX for swimming), so the weather service data is often useless.
Recommendation #3: Put temp and humidity sensors into the pod, use those for Nix index.
Recommendation #4: implement the feature I described above for people to actually use the NIX index. I will note, you don’t need this index at all, because you can also search through the activities for temp and humidity data without a combined index. But you do you.
Connection Issues
As for the bluetooth connection between the app and the pod, I almost rage quit many times. I was going to make a large part of this review about this in particular. The app would say it was connected, and then when I started an activity, it would say I need to connect. Then I would do this ridiculous cycle of connecting this before that and the other way around like a monkey trying random things to get a treat. This happened many times.
However, the day before I started writing this, I got an email about an update that supposedly fixes all that. I ran out of patches for this article, so I have not been able to test this. So I will not send Nix to the fiery depths of Hell in this article. Maybe the next one, if I repeat this study and it is still bad. Let’s assume the connection issues are all gone.
I will say, that due to these connection issues I lost an activity. I contacted support. Their support treated me very well, respectfully, my data was in their servers, and they were able to regenerate the activity in my phone. That’s the only reason that I was able to keep my cool later down the road to this article.
General utility of the software
To be blunt, I don’t see the software as very useful for a triathlete. I wear two watches so I don’t have to flip between screens and I get all the data I want from very quick glimpses. I have a cycle computer for my bike that tells me what I want. I do not want another device or screen telling me stuff I should already have a feel for.
The utility of this device is to put it on, do your training, come back and see what you got. After doing this a while, you can get good feel of the water and electrolyte loss rates and utilize those for future activities. Which is how I used it. It DOES work this way if you want.
I would connect, start the activity in the app, go do my training, then come back and wait for it to connect and sync again, and then it takes a few minutes for all the data to arrive. This worked really well all but one time.
Recommendation #5: Be a sensor, not an app. Make your sensor BLE (maybe ANT+ also? it must already be BLE), and work with Garmin, apple, Coros, etc to be something that those devices record in addition to all the other stuff they record (pedal power, cadence, etc etc etc). The last email from you suggested Training Peaks integration. That’s good, do that.
Recommendation #6: Be a platform not an app. You are already recording all our data, let us access it with cool insights online. Tell us what the population of NIX users looks like with histograms of people’s waterloss and electrolyte concentrations are. Give me personal insights about my sweat profile there. Put a store on this website to recommend hydration products that match our sweat profiles. Make there be a reason to go to your website. Otherwise, why did I make an account?
The Data
Well, this is really where the rubber meets the road, isn’t it? Everything above is fixable. But if the data is bad, well that is going to be a problem.
So lets start with my protocol so you know how I got my data. Here it is:
- Drink and eat everything I plan to do before my activity
- Put on the patch and pod
- Get naked
- Weigh myself and record the weight in a spreadsheet
- Start Nix app
- Go train, keep track of water intake
- Upon return, wait a bit until sweating stops
- Get naked
- take a shower (protecting pod)
- Dry off and weigh myself again, stop the activity in the Nix app.
Uhhh… somewhere before step 6, there should have been a note that I got dressed again and I generally don’t train naked. Also, it is critical that you do not pee between step 4 and step 10.
After this protocol, I would go into Training peaks and grab more data and record it (this is exactly the sort of thing Nix could be doing for me on their platform after granting access to my training peaks data). This way I track temperature, heart rate, and other things from my activtity and Nix data. So you get something that looks like this:
The presumption here is that my measured weight changes are all from water loss. So I need to keep track of how much water went in also. This way I can actually get a water loss rate. But why does it change?
Temperature and humidity and heart rate. If I walk or run for 30 minutes in the exact same weather conditions, I will have different water loss rates. As long as I use my Garmin gadgets I will get measurements of temperature (Apple uses weather service to report temp, which is almost always different than what I actually experience, another note to Nix about using that reliably).
The pink rows are where I used Nix sensors to measure my waterloss and electrolyte. The yellow columns are calculated data. Let’s focus on water loss for now. Also, please excuse the use of Oz, Km, mg, and Farenheight (I actually record in Celsius). Check this out in the Nix app:
Milligrams of salt per ounce of water huh? Not milligrams per liter? Or milligrams per milliliter? Anyway, the engineer in me says marketing made a decision here to get numbers between 1 and 100. Anyway, this made me have weird units also in some of my columns.
So? Whats the data look like? Well there are 2 factors to consider, Temperature and Heart rate. So lets look at those first. I removed one outlier datapoint, that gave ridiculously low numbers. Bad patch? Bad connetion? Who knows?
That isn’t awesome. The Nix waterloss rates (measured in Oz) are inversely correlated to the real water loss rates. It seems like there is a correlation at lower temperatures (training indoors) and another correlation at higher temperatures. Again, this is likely because they are using temperature data from the weather service. Let’s look at heart rate.
Better, but not great. You can see that there is a time with large waters loss (Nix and using scale) that is skewing the data at low heart rates upward for Nix. So, what we need to do is to plot the difference between Nix and measured, and then try to combine temp and heart rate together. For this, I normalized the temp and heart rate around the minimum measured numbers. This gives me numbers that are higher than 1, and when multiplied together, it will accentuate the difference between lower number and higher numbers for those metrics. What we see is interesting (to me).
So, remember, the x-axis is meaningless to humans. its just a number that gets much larger when both Temp and HR are high. Let’s call this axis “Intensity” (because training is more painful at high heart rates and high temps). The y-axis is the difference between the Nix and scale measurements. If its negative, the Nix recorded a higher number.
We clearly see two groups here. When the intensity is low, Nix seems to over estimate water loss. When intensity is high, Nix underestimates it. Well let’s eliminate the amount of time spent on an activity, and just compare rates:
Huh. Two groups again. What separates those groups? Let’s look at the difference between the rates and see if we can see.
I think we can say that the Nix hydration sensor needs to measure local temperature AND heart rate to accurately measure waterloss. Looks like the previously defined ‘intensity’ again separates them.
I think they know this. Because when you go for a training session, it asks you how fast you are going to go. I measure my performance when riding by power, not speed. When I do any activity, there are usually intervals that take me from Zone 2 to 4 or 5 sometimes. Point being, I don’t know my average speed until the session is complete.
Recommendation #7: Link waterloss data to local temperature and heart rate. You already do a lot of correlations to get a waterloss model. Each correlation has the potential to make the result worse. Why not calibrate the patch to the person by having them do one trial run with the protocol above?
Recommendation #8: At least lose the requirement that the athlete needs to know their pace for the activity they are about to do. Use cumulative data to suggest hydration needs, and course correct as they use it.
What about electrolytes?
Looks like the electrolyte measurements are also affected by intensity, but temperature is the bigger factor there. This is really good. In your “Nix Science” section on your website there is a temperature sweep from 18 degrees to 30 degrees (I assume this is celsius?). that correlation looks good. So while your lab models seem good, somehow the results in real life need work, I suspect that this is because you dont measure local temp.
Conclusion
For waterloss, it looks like you can expect an answer that is up to 100% off at low intensities, while lower error rates can be found at higher intensities. The average error rate is 17%, but that really doesn’t describe the variability between low and high intensity. At high intensities, the error averages to 15%, and at low intensities the water loss error averages to 56%.
Considering an Ironman is primarily run in Zone 2, The data gathered from the Nix is a poor predictor of race day needs, especially if it isn't too hot.
The electrolyte measurements are better, and I feel pretty comfortable using them. The average (for me) measurement for high intensity is 58.5 mg/Oz (using those weirdo Nix units), while at low intensities it’s 52.4. I am happy to use 55mg/Oz as a predictor for race day. However, it should be noted that I never got this done professionally to compare this number to. It may have some offset I don’t know about.
One final note, these patches are expensive. They are 5 dollars each. You get a pack of 5 to start. But after that you are on your own with the the ink jet model of milking users as much as possible. This will make it so that as soon as someone is comfortable with their hydration needs, they will immediately stop using them. If I didn’t want to write this article I would have stopped after the first 5, sold the pod to someone else.
Recommendation #8: Make the patches cheaper. you want people to keep building a profile. You want to provide data to them about long term effects such as how the seasons change, maybe changes in diet, etc. But for you to keep providing that information, you need to make the patches way cheaper. Like less than a dollar sort of cheaper.