[Demo Recap] Microsoft & DH2i: Analyzing ECG Data for Heartbeat Categorization

In October, DH2i CTO OJ Ngo had the pleasure of joining Microsoft Program Manager Vasiya Krishnan and Azure IoT Sr PM & Developer Advocate Pamela Cortez for an IoT Deep Dive webinar on our joint IoT security solution. In this session, Vasiya and OJ provided a demo to showcase the end-to-end data security offered when combining Azure SQL Edge with DxOdyssey for IoT. For a rundown of the demo, check out the highlights below.

Demo Overview

This demo outline walks through the process of monitoring a patient’s ECG (electrocardiogram) data to provide their primary care physician with vital information about the patient’s heart health. The patient doesn’t have access to their physician – instead, they wear a portable ECG monitor that needs to connect to their physician’s remote PC to allow the physician to access the patient’s data.

Creating Patient Data

The patient is located in a place with sporadic internet connectivity in which they are not always able to send data to the cloud. The portable ECG monitor is set up to constantly receive data from the sensors on the device and then retain that data to Kafka (an open-source stream-process software) for processing.

Patient Data Generating from Edge Device Sensors

 

Communicating Patient Data with Azure SQL Edge

Streaming the data from all sensors to a single device is extremely important for decision making. In this case, the physician will take into account multiple data points from the generated data. The data from Kafka will be retained to Azure SQL Edge, which is deployed remotely as a stand-alone container on the portable ECG monitor. The next step is to form the connection between the ECG monitor and the physician’s remote PC with DxOdyssey for IoT.

Deploying DxOdyssey for IoT

The DxOdyssey for IoT module can be found in the Azure Marketplace in the IoT Hub where new IoT Edge modules can be set. The different environment variables must be set up next to proceed. Once deployed, the DxOdyssey Management Console can be used to create and manage the dynamic tunnels.

DxOdyssey Management Console

The Management Console provides all relevant networking details. The image below displays the details for the deployed DxOdyssey edge module. The PC and the edge device running in Azure are sitting behind private networks. Both are running DxOdyssey for IoT and can securely connect over the public internet via DxOdyssey micro-tunnels once created.

DxOdyssey Edge Module Networking Details

 

Creating Secure Tunnels with DxOdyssey for IoT

The “Add Tunnel” option can be found within the “Tunnel Manager” menu option. Once the dialog box opens, the tunnel name and the target that is seeking access must be defined. For this demo, the gateway is the DxOdyssey edge module (DXOEDGE1), the target Host/IP is the SQL Edge container name, and the target port is the port that SQL Edge is listening on.

Tunnel Management Dialog Box

The next step is to define the origin of who is allowed to connect through this dynamic tunnel. DxOdyssey for IoT allows users to create micro-tunnels for specific applications.  In this case, only the PC (VASIYA-MSFT) needs access to the SQL Edge application on the edge device in Azure. Once the connection has been established, the data generated from the ECG monitor can be securely forwarded to SQL Edge on the physician’s PC.

Secure Tunnel Details

 

Analyzing Patient Data for a Solution

After accessing the data in Azure SQL Edge, it can be pushed to the cloud for ML model training. Azure SQL Edge offers various built-in ML capabilities that allow users to build models on the cloud or datacenter and then deploy it to the edge. Once the ML model training is complete, it can be deployed to SQL Edge via tunneling to provide the physician with the necessary data.   

For this use-case, the model took into account the latest 10 data points received from the edge device sensors. Looking at the generated data for ECG Classification, all data points were either a 1 or 0. Heartbeat classification consists of 5 stages from 0 to 4, with 4 being the most fatal state. Given the results, this patient is well within the normal range and does not need immediate attention from the physician. The ability to access, process, and analyze data in real-time anywhere enables physicians to more closely monitor their patients and potentially prevent life-threatening illnesses.

Heartbeat Classification Results

 

To view the full webinar and demo, follow this link: https://aka.ms/deepdive/Azure-SQL-Edge-DxOdyssey

Ready to try DxOdyssey for IoT out for yourself? Access the offer in the Azure Marketplace.

Connor Cox