Has anyone here used the EmotiBit for a self-tracking project? It is entirely open-source and collects a lot more data than most devices. It’s a bit pricey, though, and since my primary concern is tracking heart rate, I might opt for a cheaper, ECG-based device like a Polar H10. At the same time, it would be awesome to write some of my own software to parse its data into metrics about what I’m doing at different points in the day, how much I exercise, how well I sleep, etc.
Would love to know if anyone has any experience with the EmotiBit or similar open-source devices.
I was thinking about buying it multiple times, but each time i was not satisfied by the specs. Here are some points
- battery doesnt seems to be great, it seems device needs to be charged every 6-8 hours.
- sampling frequency also doesnt look good if you want HRV. by default it set to 25Hz and as far as i know there is firmware with 100Hz, but i not sure if it can be increased to pass my requirements (256Hz minimum for HRV).
- after my own experiments with other devices i came to view that PPG isnt good for HRV metrics when person is in motion and sometimes even in stable position (like sitting).
- device have EDA sensor. I own a research grade EDA sensor, but still was not able to get long term data interpretation in a meaningful way.
- temperature sensor might be fine, but when measuring skin temperature you need to think where you want to put sensor and what you want to measure. Usually you want it to be as small as iButton to place it everywhere you want, but EmotiBit is not small and some locations will be not possible / not comfortable.
If you look for short recording sessions it may work, but i’m sceptical about long term use with single device. You may need two to change them when battery runs out.
You can look what device can capture
As you can see they advertise some metrics like Respiration, Hydration, SPO2 or Body Temperature. But skin temp is not body temp and it almost impossible to get proxy to it with just 1 temp sensor placed on a skin, and 0.001C resolution is not same as 0.001C accuracy))) Also i dont beleive that you can get good SPO2 readings from wrist ppg with reflective technology in real life conditions…
I dont know what they mean as Health from Humidity and Temperature section (just advertising?)
But i dont say that device is bad, it may be good for some specific use-cases.
Polar H10 mostly measures HR and can be used to measure HRV in a short sessions. It doesnt have a lot of sensors as Emotibit, but for what it have it offers better data quality for sure.
From my perspective it is better to focus on single biosignal, get high quality data and be able to interpret it instead of having ton of sensors with average quality which will be not an easy task to interpret.
About the battery–I believe any battery compatible with an Adafruit feather would work, so buying a battery rated for higher mAh to last longer on a single charge should be possible. I think your other concerns are very valid, though, and I didn’t know that about the sampling frequency for HRV, so thank you.
I know Apple Watches use PPG to measure HR and HRV unless you are using the ECG app; if you have any experience with Apple Watches, would you consider them accurate enough for HR and HRV when moving? I’m trying to get a sense of what is possible with PPG, given strong software.
Thank you again.
I dont have enough experience with Apple Watches PPG. They seems to have very good algorhythms to estimate heart rate from PPG, but i never saw long term HRV data from Apple Watch. The data from Apple UI seems to have low resolution (few minutes), but maybe somewhere they have raw data with high resolution.
HRV calculation have higher requirements for data quality, often you can get good HR but HRV will be inaccurate. This happens because HR is just count of RR intervals per minute and its not a big deal if some beats misaligned. In contrast, HRV is a set of metrics which reflect intervals between R peaks and even single misaligned peak will distort results. HRV often focuses in high-frequency range as measure of parasymphathetic system and this requires high sampling frequency, for good HRV i think 256Hz is a minimum to start with.
Bigger battery means less comfort, imagine smartwatch with 100g+ of weight. If current battery lasts 6-8 hours then you will need x3-x4 battery for a whole day, and after you take it off it will have some time to charge and you will have gaps in your data (but this doesnt matter if you dont need a full 24/7 setup). Another option is to lower sampling rate, but this will result in lower data quality.
Before buying any devices it is good to start with formulating specific questions you want to answer, what data is required for answer and what decisions you will make if you have answers.