Fetch raw sensor data from several fitbits and on-board processing of data

[ie4721]

Hey,

I have a few questions to hardware and software folks. This is an academic project: 

Project overview:
I'm building a Digital Twin system for the Emergency Room (ER) Department at the University and City Hospitals to holistically study all parameters contributing to long wait-times for patients to receive care as well as associated contributing factors, including overworked, understaffed ERs.

We're using non-invasive wearables (either built from scratch or off-shelf, e.g. apple watch ) to fetch health status data into a cloud service (e.g. Azure/Google Cloud). If I'm building the wearables myself, I'll be assembling the sensors into a microcontroller and run embedded machine learning (e.g. anomaly detection / forecasting) on the device itself (TinyML), then upload the data to the cloud via cellular connectivity.

Questions:
- Aside from API access, is it possible to have access to the device itself to accumulate data and/or run specific ML models? i.e. fetch raw data in near-real time and do custom processing on the device itself or in the cloud. 
- Some wearables have data collection modes, i.e. collecting data continuously during workout and periodically (~10mins) during rest. Is it possible to modify modes?
- Is it possible to turn each wearable into a network node (i.e. rather than fetch data via API into a central gateway)?
- I didn't see a digital twin application through healthsolutions.fitbit. Is there a similar solution to what we're trying to build?

Many thanks!

PS: I can provide more info about the project and system architecture if you reach out to me privately. Not comfortable sharing details on a public forum...

[Gordon-C]

Hi @ie4721 

Raw sensor data is available through our smart watches using the Device SDK.   I've moved your post to the SDK forums.   Hopefully, someone here can answer your questions.

Gordon