My name is Justin and I am an engineer for a health technology startup in the UK. We are working on a few digital health solutions which naturally involve a lot of research. For a particular research project we realised we required a convenient manner in which to collect ECG and PPG signals. This lead to developing a device to resolve the requirements of data collection. We now have a wearable datalogger that can capture ECG, PPG and a 9 DOF IMU all embedded with Wi-Fi and BLE. We also added an extra I2C port for additional sensing possibilities. It has been working really well for us as it is synced up to our internal metadata service which allows quick access to clean, long term physiological data for analysis and insights.
I have come here to find out if anyone has any experience in these types of signals and/ or would be interested in continuous monitoring of this nature. We would really appreciate any feedback and are keen to learn about possible use cases for a potential product that could be developed.
Thank you for posting this here. The short answer is: yes.
If you browse around in the forum you will see some very good discussions of projects using ECK data. For QS projects, the PPG signal is most frequently encountered in blood oximetry and HR detection. I haven’t seen the signal used for other purposes but others may know more.
An unsolved problem, as far as I know as a user of some of these devices is that while HR and related data types, like HRV and RHR, are available in convenient export formats for analysis, full ECG signals generally are not. Common devices, such as the Kardia that I own, only provide visual representation and PDF export. I’m about the try the Zio patch and as far as I can tell this also only provides PDF reports.
That is great to hear - I will certainly do some more digging through the various topics to see what I can find. We have designed this device with research in mind so the process of capturing data continuously and conveniently were the primary design requirements.
With that said, we have catered for the data to be stored on the SD card in .csv format, which is accessible from the device. It is also capable of easy upload to our servers that (could) be configured for access to download the data through a web dashboard (our data team uses API calls at the moment but we could potentially create a simple UI). So as far as that unsolved problem goes - we have developed a device that potentially fills that specific gap. At the moment we are just using it internally but it would be great to see if there is any interest in the device for research and analysis purposes in the QS community.
We are most intrigued to learn whether there is a use for continuous monitoring of these signals (ECG, PPG, motion). And also if there are any ideas that come to mind about adaptations that could make it more useful (different signals, specific research to look at etc.). Very open to criticisms and all types of feedback!
Some basic specifications:
Onboard Sensor Outputs
PPG Waveform
ECG Waveform
9 DoF Absolute Orientation
Number of Channels:
Analog (2)
Very interesting indeed. One high level bit of advice is that it’s going to take a lot of ecosystem support to make these signals meaningful to users/participants. I definitely think there are at least a few technically competent users on this forum who would be able to experiment immediately, but for broader use the details of how the data is saved, shared, visualized, and integrated into a step by step process of learning are crucially important. There has been a long series of promising sensing tools that disappeared due to lack of this ecosystem support.
I do not use contec because it is unwieldy and only a better than smart watch PPG. I want to track as much as possible including ECG so I am interested. Do you include oxygen saturation since you already have ppg? Many newer chinese smartbands offer blood pressure monitoring with ppg. Is that real? Would it be easier with ECG?
@Agaricus Absolutely - I think that there is no doubt surrounding the effort(s) required to create a supporting ecosystem for a usable product. At this stage it is really a byproduct of an internal research project with the potential of becoming something useful. What are the most common forms of data interrogation for QS users? Quite a broad question but would be useful to know what level to cater for in this community vs. what research tool usability expectations are?
I think the intention with coming here is to test the waters to find out if there is any interest in these signal types and or what they could be used for. Some ideas that are within the scope of discussion and with viable development potential are:
Heart Rate
Heart Rate Variability
Respiration Rate
Respiratory Rate Variability
Pulse Transit Time
Perfusion Index
Pulse Wave Velocity
R-R intervals
Continuous Blood Pressure??
Continuous Motion Tracking
@rain8dome9 Thanks for the feedback on the Contec device - will have a look at them. Which model are you referring to out of interest? Great that you are interested in tracking ECG. Are there any aspects of the ECG signal that you are particularly interested in? We don’t have oxygen saturation at the moment. The analog PPG signal on this device is composed of a single source wavelength of light (IR) so it is not currently possible to do SpO2. There is literature surrounding continuous blood pressure with PPG - I wish I could tell you if it was real. What more can you tell me about your interest in it?
There is a watch-finger PPG model. The more your product can do the more I will want it. I have no specific goals besides More Data that is affordable and easy to get. I do want software that can do the initial signal processing but it could easily be FOSS.
Definitely one approach is to provide the device with a design optimized for personal control of data. So, the sensor stack plus thoughtful usability/openness tradeoffs involving on device storage, battery life, data transfer, firmware, etc. There are so many dials to turn that its hard to know what’s going to count at the beginning, but if you can do an inexpensive open solution that provides pulse detection, plus some open tools for outputting various HR related data types, you might be able to build a user base of technically proficient users who can then contribute to the ecosystem. Our friends at Open Humans have been building a platform for collecting HR related data and exploring it in relation to symptoms of infection (Quantified Flu). They are doing this via user permissions to query Fitbit, Oura, and Apple Health for HR data. That ongoing project might be a good thing to look at to see how HR data is being used in the current COVID-19 scenario. But there have been a lot of creative projects. Here’s an unrelated one about HRV and sleep.
@rain8dome9 Can you elaborate on what level of datasets you are interested in? Do you use any dataset interfaces currently (software, web, other)? If so, what are the features that you are looking for in an interface? Do you have any examples of datasets that you commonly use? What aspects of the data acquisition pipeline do you think are the most important -> device/ tool, signal processing, accessibility to data, user interface?
@Agaricus Some great thoughts there. Are you aware of any devices that have come close to offering a good trade off between usability and openness? What types of platforms do you make use of for data interrogation? How many “technically proficient” users would you guess there are in the QS community? Do you think that the easier the interface & usability of the device the greater interest there would be from the community? Do you have any idea of what the most commonly tracked physiological signals are amongst QS? Really interesting to see what is happening at the Quantified Flu - are you aware of any major challenges they are facing? Happy to go and find out myself too if you don’t know - just thought I might ask. Where could I find out about some quantified self thought leaders? It would be really interesting to hear about what trends people are interested in at the moment and who are people listening to? Lots of questions - please answer what you are comfortable with.
Hi Justin - I appreciate these questions, they are very good one, but impossible to answer briefly. @Steven_Jonas and @jakobeglarsen have done an analysis of some of the topics of talks archived in the QS Show&Tell library. Keep in mind that this is just a list of tags on the topics from a limited sample. I would rely less on the absolute numbers or the rank order than on the topics themselves. They give you a sense of the breadth. (I’m not sure why HR/cardiovascular and fitness appear more than once, I think it is an artifact of mistakes with the tags, the numbers should probably be added together.) In any case you can see that HR related projects are very popular.
The best way to learn more about Quantified Flu is to sign up and begin tracking your HR, body temp, and symptoms (or lack of them). https://quantifiedflu.org/
Does 26 on cog seem like a lot or a little? Unfortunately our web archive search by topics isn’t working very well. But I ran the search on our Filemaker database of archived talks, and here’s what I found. If you search for these by name or title in the web page they should show up.
Hey guys - thanks for the responses here. I must have not seen the notifications in my mail so apologies for taking a while to get back to you. Appreciate the insight on the topic depth and will certainly spend some more time digging through the forum and watching talks. I guess that my questions probably need a bit more of a focused conversation with some people in the know.
Thanks for the direction! Will keep in touch with regards to progress and new ideas back here.
