| I worked on a research project for several years (ending around 2018) using smartwatches to record a variety of physiological measures like stress. We were always looking for an open platform to use, and tried a variety of Android Wear watches and the Apple watch, but found little with the features we wanted. Microsoft used to make a smartwatch that had open access to almost all the sensors via Bluetooth, but they stopped making them some years ago. So we designed our own. All the design info and firmware source is available here: https://github.com/AmuletGroup/amulet-project This let us build a custom operating system that could run multiple apps and include external sensors via Bluetooth plus a few internal sensors plus a link to a smartphone (and it's sensors). Near the end of the project I had a bunch of them (125 I think) made by an electronics assembly firm. If you want to build your own those files might be a starting point, but I'd also add the warning that many of the components, particularly the nRF51822 radio/MCU are likely no longer available in that exact version (Nordic changes their chip design often) which also means the firmware would need porting to the latest SDK. The touch interface chip is now obsolete. To get 125 built I had to pre-buy all the critical components we needed (some of them off eBay) and send them to the manufacturer. The e-ink LCD panels were hard to find also. You might be interested in this: https://hackaday.com/2019/02/20/custom-firmware-for-cheap-fi... That Arduino IDE generated firmware will run on a variety of cheap fitness trackers from China (the ones that use an nRF52832 radio/MCU). I haven't been active in this area for a few years so there may be new products available now. I see there are a number of self-built smartwatch projects on HackADay: https://hackaday.com/?s=smartwatch I'd also warn that getting reliable, accurate medical sensor data from a smartwatch or body worn medical sensor is not a solved problem. Unlike traditional medical sensors used next to a hospital bed a mobile sensor is subject to a lot of motion and environmental changes which messes with the readings. It's difficult to keep a sensor attached to one specific place on a human body. The more advanced sensors you'd like such as blood pressure are not available yet. Heart rate variability in Garmin's is usable, but wrist or body motion can throw it off. A chest strap such as a Polar H7 will give you a much better heart rate and HRV reading, but many people find them uncomfortable to wear. Blood oxygen readings can be fairly accurate when the person is not moving and the strap is tight, but will vary wildly once the person is motion. Be prepared to have to filter the data (e.g., throw out sensor data when the accelerometer indicates motion). Do a search for the new devices promising blood pressure (Samsung I think) or blood glucose monitoring (Apple). One of those might fit your needs when they become available. Beyond that have a look at medical web sites to see what hospitals are using (although those devices tend to require a prescription and are usually only sold to hospitals). You might find some leads here: https://www.mobihealthnews.com/ There are some research projects in this area too, here's one of the big ones: https://md2k.org/ led by Santosh Kumar: https://www.memphis.edu/cs/santosh-kumar/mhealth-systems-lab... Deborah Estrin has done a lot of excellent research on mobile health sensing as well: https://en.wikipedia.org/wiki/Deborah_Estrin
https://destrin.tech.cornell.edu/ Her Open mHealth project might be useful for making sense of the data you collect: https://www.openmhealth.org/ You might find some leads from Tanzeem Choudhury's work also: https://pac.cs.cornell.edu/ though her work tends to use smartphones.
Good luck! |
This device is FDA approved to what you described (not blood glucose yet--but is technically capable): Biobeat Watch BB-613: https://www.bio-beat.com/copy-of-clinical-trials-and-researc...
It is Israeli technology and they do excel at things like photoplethysmography (PPG) and coulometry, from a historical standpoint.
But, the Empatica E4 is 100% capable of doing what you described above. You just have to use Lab Streaming Layer to interface with whatever you are programming in, such as Python. You basically reimplement prior research papers in your own code. However, it does not have an SpO2 sensor.