Hackathon IoT code to support Applied Behavior Analysis

This scenario is targetted for creating a simple “clicker” that can help record/count events vital to helping in ABA therapy. At the core, it is about having a dedicated device to record clicks and automatically share the information in a cloud. Having a dedicated device with a physical button(s) removes all the distractions of an app on a shared device.

Button → Microcontroller (NodeMCU) → Cloud (Azure Iot Hub) → Azure Function (moves IotHub to CosmosDB)

There are a few basics one layer deeper. The button needs to be wired and housed in a case with the microcontroller and a battery. The microcontroller needs to be connected to a wireless network and have secure access. The cloud needs to supply an endpoint to capture the data for later analysis.

For the intial version, we are looking at the following.

Button : Any simple momentary button switch. This can be any hardware to capture an event.

NodeMCU: https://www.nodemcu.com/ This controller is fully open source, fairly easy to load/program, has very low power consumption, and is inexpensive. We can power it for the hackathon using any simple USB battery pack. We will leverage the Arduino IDE to ease in programming. https://www.arduino.cc/

Physical Case: While the proof of concept might be shown using a cardboard box or tupperware container, the hackathon will start the design of the custom case, which can be 3D printed. The designs may use any modelling software. We will start with FreeCAD or Blender (https://www.blender.org/) The native blender files as well as STL files for ease in 3D printing will be shared in the repository.

MQTT Protocol: https://mqtt.org/ MQTT is becoming the standard for IoT devices. Most Cloud providers have a platform for receiving these messages. It is lightweight and allows for easy asynchronous communication streams. Azure has a mechanism which can be used in the hackathon.

Data Analysis: TDB.

Infrastructure in Azure

The infrastructure is running as Serverless in Azure and is deployed and managed with Terraform. The following items are deployed. Refer to the code in the terraform directory

• CosmosDB
• App Service
• Function App
• Application Insights
• IoT Central
• IoT Hub App
• Storage account

Backend APIs

Azure Functions is exposing APIs for below. The APIs are a mix of .NET Core and Python. The Functions connect to CosmosDB on the backend and the database connection/key is stored as a secure application setting inside the function. The code can be found in the backend directory

• GetButtonEventsForSession
• GetSessionsForPatient
• GetButtonSemantics
• GetPatients
• SetSessionDetails
• StartSession
• EndSession

Outline of flow

• Frontend UI makes an HTTPS request- > backend Azure function endpoint -> read/write data to CosmosDB

Next steps / TODO / Nice to haves

• Consider PII management for patient data (encryption at rest, in transit, etc.)
• Authentication between frontend and backend, for example, JSON Web Tokens (JWT)
• Therapist/Client Authentication with Application
• Administration of Therapists / Patients / etc.
• Adding/Removing Devices
• ADA Accessibility on devices o Haptic Feedback?
• Light(s) for status on device