> Now we're doing everything we can to move computation to the edge. In an ideal world, the cloud wouldn't store or process anything - just receive already-analyzed, privacy-preserving results straight from the device.
I appreciate moving away from cloud for personal health data like this.
Can you clarify why the ideal world you hint at would have the device sending any data to you?
Devices sending data that is anonymized, encrypted, and signed by the device is a must-have for some medical studies.
For example, imagine a medical study that looks at heart rate variability versus an intervention. The data analysts won't need to know each patient's name or email address, but will need to know each patient's heart rate variability when you're having the intervention. The study may span many physical locations, such as at multiple medical providers across a country.
Interesting point about anonymized data! I've been thinking about how sensitive health information can actually make or break a study. The idea of sending only encrypted, signed data sounds perfect for that. I mean, I wonder how many studies are limited just because researchers can’t get the data they need without compromising privacy.
I've seen this with other health apps too – the balance between usability and security is tough! It reminds me of the push for end-to-end encryption in messaging apps. Like, we have the tech to keep things private, but how do we still provide value to researchers who really need access to those anonymized datasets for insights?
I'd be curious if there’s been any pushback from the medical community on this approach. I can imagine some would argue that the more security layers there are, the harder it may be to analyze trends effectively. But maybe that's just the growing pains of tech in healthcare?
for regular users, since the device itself doesn't have any interface, we need to send at least some data so they can see it somewhere. the natural place is the phone, but not all of our professional clients liked viewing long ECG recordings or detailed metrics on a small screen, so we built cloud access mostly for convenience .
(to be clear - if a developer wants, they don't need to send anything to us)
I've been dealing with some cardiac issues that were brought on / exasperated by a covid infection, and it's been challenging finding solutions that let me monitor my own biometrics while hiking without sending anything to the cloud. It's shocking that it's a non-trivial endeavor to pair a medical grade wearable with a smartphone and get differentiated granular alarms / alerts for user defined events (BPM within user defined bands, blood oxygen levels below a certain threshhold, etc..)
If your python SDKs work wirelessly, I'd seriously consider creating a RPI based system to do those things and be able to leave the smartphone at home while I hike.
Hey, can you join our Discord: https://discord.gg/sPay3Xm? Reach out to me personally (@guzik1) - happy to help you with that as we've already had a few people who integrated with rpi, so I can share some tips and examples.
When you're ready with Aidlab 2, can you contact me? I work for a national health service and I'm keen to learn more, buy, and generate some public anonymized data sets. joel@joelparkerhenderson.com. Thanks and good work! <3
Any information on how comfortable the strap is? I am wearing a Garmin HRM Pro for one hour a day during workouts and it is not very comfy. I know a lot of athletes are moving to way less precise optical hand straps just because of the comfort issues with chest straps. I would not wear a chest strap for longer periods of time, unless I absolutely had to.
sadly, comfort for chest straps compared to hand straps is a known issue and ours is definitely no different. Wve done a bunch of tests, tried different materials/custom solutions, and honestly we're still clueless how to make it significantly better (if anyone here works in textiles or wearable fabrics, I'd love to connect). So yeah, if wearing your Garmin for more than an hour already feels uncomfortable, ours probably won't be much better in that regard
I watch "Quantified Scientist" (https://www.robterhorst.com/), in which Rob compares various watches and smartbands on how they measure sleep and heart rate - against golden standards.
By any chance, were you able to talk with him so he can measure it?
Thanks! Yep, the interface for a shell is available here: aidlab.com/developer/debug we are using Jquery Terminal + Web Bluetooth (sadly, I think it's not under active development anymore)
and yeah, when we started years ago, it felt natural that the next step would be to measure glucose from blood but the truth is with the current state of science, it's still not possible to do that 100% non-invasively.
> Now we're doing everything we can to move computation to the edge. In an ideal world, the cloud wouldn't store or process anything - just receive already-analyzed, privacy-preserving results straight from the device.
I appreciate moving away from cloud for personal health data like this.
Can you clarify why the ideal world you hint at would have the device sending any data to you?
Devices sending data that is anonymized, encrypted, and signed by the device is a must-have for some medical studies.
For example, imagine a medical study that looks at heart rate variability versus an intervention. The data analysts won't need to know each patient's name or email address, but will need to know each patient's heart rate variability when you're having the intervention. The study may span many physical locations, such as at multiple medical providers across a country.
Interesting point about anonymized data! I've been thinking about how sensitive health information can actually make or break a study. The idea of sending only encrypted, signed data sounds perfect for that. I mean, I wonder how many studies are limited just because researchers can’t get the data they need without compromising privacy.
I've seen this with other health apps too – the balance between usability and security is tough! It reminds me of the push for end-to-end encryption in messaging apps. Like, we have the tech to keep things private, but how do we still provide value to researchers who really need access to those anonymized datasets for insights?
I'd be curious if there’s been any pushback from the medical community on this approach. I can imagine some would argue that the more security layers there are, the harder it may be to analyze trends effectively. But maybe that's just the growing pains of tech in healthcare?
Ditto on this! I've avoided fitness trackers so far because I don't want any cloud aka my data on someone else's computer
for regular users, since the device itself doesn't have any interface, we need to send at least some data so they can see it somewhere. the natural place is the phone, but not all of our professional clients liked viewing long ECG recordings or detailed metrics on a small screen, so we built cloud access mostly for convenience .
(to be clear - if a developer wants, they don't need to send anything to us)
I thoroughly applaud your approach.
I've been dealing with some cardiac issues that were brought on / exasperated by a covid infection, and it's been challenging finding solutions that let me monitor my own biometrics while hiking without sending anything to the cloud. It's shocking that it's a non-trivial endeavor to pair a medical grade wearable with a smartphone and get differentiated granular alarms / alerts for user defined events (BPM within user defined bands, blood oxygen levels below a certain threshhold, etc..)
If your python SDKs work wirelessly, I'd seriously consider creating a RPI based system to do those things and be able to leave the smartphone at home while I hike.
Hey, can you join our Discord: https://discord.gg/sPay3Xm? Reach out to me personally (@guzik1) - happy to help you with that as we've already had a few people who integrated with rpi, so I can share some tips and examples.
When you're ready with Aidlab 2, can you contact me? I work for a national health service and I'm keen to learn more, buy, and generate some public anonymized data sets. joel@joelparkerhenderson.com. Thanks and good work! <3
Any information on how comfortable the strap is? I am wearing a Garmin HRM Pro for one hour a day during workouts and it is not very comfy. I know a lot of athletes are moving to way less precise optical hand straps just because of the comfort issues with chest straps. I would not wear a chest strap for longer periods of time, unless I absolutely had to.
sadly, comfort for chest straps compared to hand straps is a known issue and ours is definitely no different. Wve done a bunch of tests, tried different materials/custom solutions, and honestly we're still clueless how to make it significantly better (if anyone here works in textiles or wearable fabrics, I'd love to connect). So yeah, if wearing your Garmin for more than an hour already feels uncomfortable, ours probably won't be much better in that regard
I watch "Quantified Scientist" (https://www.robterhorst.com/), in which Rob compares various watches and smartbands on how they measure sleep and heart rate - against golden standards.
By any chance, were you able to talk with him so he can measure it?
Hey there,
We haven’t spoken with Rob yet but that is a great suggestion. Thanks for the tip!
Nice idea. Wonder if you can use web Bluetooth to connect a web page directly to it?
Having glucose would be cool too.
Thanks! Yep, the interface for a shell is available here: aidlab.com/developer/debug we are using Jquery Terminal + Web Bluetooth (sadly, I think it's not under active development anymore)
and yeah, when we started years ago, it felt natural that the next step would be to measure glucose from blood but the truth is with the current state of science, it's still not possible to do that 100% non-invasively.
Very commendable approach!
Are you using Movesense as the chest movement sensor per chance? I've been looking into breathing rate lately, but haven't made the jump just yet.
Thank you! We've integrated a different chip for measuring chest movements
Plans for blood glucose?
we don't have any plans right now to build a blood glucose sensor (although we already support a few external sensors - but not glucose monitors yet)