I am struggling to find open source device manufacturers to solve a critical medical use case.
Do you have anyone in your network (factory/manufacturer/designer/circuits/microelectronics expert) who would know who could help build an open source vitals measurement platform?
For example, we need SpO2, blood pressure, height, weight, etc, or some subset thereof, so the data can be entered into OpenMRS automatically.
After finishing my PhD I work at an academic medical center, and through this work with an NGO doing open source telehealth delivery in rural India (https://intelehealth.org).
The biggest pain point for the community health workers is manual data entry - they currently write heart rate, blood pressure, height, weight, etc on paper and then enter it into an Android app that integrates OpenMRS.
We cannot afford to use proprietary solutions, and I need a cost estimate to pitch this to a philanthropic org like the Gates foundation. The long-run goal is to solve delivery of health and intervention with open source AI and machine learning (e.g. here is an example of a system I have built used in hospitals: [1904.05342] ClinicalBERT: Modeling Clinical Notes and Predicting Hospital Readmission).
Please DM me if you need specifics.
Thanks so much! Thanks for your work on OpenEHR – I am convinced that Epic cannot solve the problems we can, especially where machine learning & AI can be used for clinical decision support in low- and middle-income countries. H/T @thomas.beale for helping articulate this vital use-case
The UK Apperta Foundation has worked on Open-eObs and I believe they were linking with device manufacturers to automate vital signs measurement. Perhaps Peter Coates (Apperta CEO) would be able to advise you.
Good luck with this important work.
Yes - you definitely want o speak to @David-Jobling and @sjm about the openEHR-based and OSS apps they have been working on which definitely cover this space.
An earlier version of the template used is at Clinical Knowledge Manager
Thanks so much David & Ian! Reached out to Peter, David, and Stuart
Fingers crossed we can find a solution here – will do my best.
Thanks Ian - have followed up.
If you are looking for support in designing the data capture devices, do let me know and I can try to connect you with some people working in the medical devices space.
Thanks so much Dileep! Yes, it would be great to connect with people in the medical devices space – my email is email@example.com.
Are you looking for existing using off-the-shelf electronics products and making data from these available in openMRS, or are you looking to make the actual electronics? The latter is possible, but it’s an area with quite a lot of patents and also required certification which makes it difficult. The easiest is to use off-the-shelf electronics that support BLE-interfaces and then make a mobile app that can upload this into openMRS or similar. That’s relatively simple.
Here’s an example of a pulse belt app (though it’s many years since I wrote this), and quite a lot of equipment supports similar interfaces. GitHub - mobilars/BLEConnect: Android app for downloading heart rate data from the Polar H7 chest belt.
Hi Jaan, we have some knowledge and expertise with ISO 11073 (point of care devices) and are currently looking at persisting vital signs data as part of an open anaesthetic record standard. Happy to help, although I have a clinical background rather than electrical engineering.
Thank you @Lars! Very cool to see your work with BLE integration (I have a Polar H7 myself
(1) I agree that off-the-shelf electronics products can currently be linked to openMRS / openEHR with Raspberry Pis and similar boards (Stuart pointed me to similar examples he has built like OpusVL Blog: HSCIC Prototyping tele-health using RaspberryPi). I am getting quotes to see what is possible in this space.
(2) That is a near-term solution. In the long-term, more open electronics need to be built such as pine64.org and linked with openEHR/openMRS. Do you know any experts in the patents and certification processes, or anyone to talk to in this area?
Thank you John! Sent you an email
I’ve just been making CE-certified equipment, which is a european standard. Not sure what’d be the requirement outside of Europe and US, or who would know. I think though that making medical sensors good enough and cheaper than the large manufacturers and knock-off chinese suppliers is difficult. I know from experience that making a good pulsox is quite difficult and one is typically better off using an OEM module. It’s a few years since I worked on it, so my guess is that there’d be some really cheap OEM-products out of china you could use. A pulsox with bluetooth is just 20USD at banggood.com. Designing and building enough would be much more expensive. If you’re thinking ECG also… well… that’s another matter…
You can do some research on Continua specifications and certified devices, but in the end they’re basically compatible with the Bluetooth Low Energy specifications. There are bluetooth-enabled pulsox, body scales, blood pressure monitors and some others. The chain is basically:
Sensor (uses bluetooth) → personal gateway (runs on phone or as you say PI) → openMRS
For personal gateway, well, I think your best bet is using Android phones which the health workers are likely to have already and will be comparable to prices of the Raspberry PI. If you’re determined to use a device other than a mobile phone, an ESP32 might work (has WIFI and Bluetooth and is very cheap). You’ll find example code for interfacing to a footpod from an ESP32 over bluetooth here GitHub - mobilars/RSC: Based on ESP32-based solution from https://github.com/imwitti/FootpodMimic (but not the api towards openMRS but that should be really simple). It’s a fun project if you have a treadmill ESP32 Bluetooth for Threadmill speed to Zwift - YouTube (but sorry, that’s off topic )
Edit: I just remembered now that this example code actually is code for emulating the sensor rather than the gateway , so if you want to do some demos with implementing sensors you could use this. It’d simply be a matter of changing the GATT-profile (bluetooth profile) to fit your sensor. The earlier code I posted is the gateway, but it’s really old Android code so not sure it works any more.