CheckMate

Azure ML Studio
Architecture

Inspiration

During Covid, there was an extreme shortage of the medical devices that are used to measure the body temperature in a non-contact way and Oximeters that can measure SPO2 levels. Usually, these two tasks are taken care of by two completely different devices, which results in the purchase of more than one device to measure health, also, these devices were a luxury in developing and undeveloped countries where cost is a huge barrier in getting a proper diagnosis. Thinking about this problem motivated me to do something which eventually became CheckMate. It is an embedded solution that leverages the power of Azure Machine Learning to solve the problem of expensive non-contact thermometers and oximeters.

What it does

The hardware device, still in the early prototype phase, takes in incoming data from the con contact IR thermopile sensor and an SPO2 sensor, the microcontroller, ESP8266 connects to wifi and hosts a very basic server that responds to the requests made by the web app, where the data of Temperature, Heart Rate, and Oxygen Saturation is displayed in an easy to understand UI. The second functionality of the web app is to take some basic information from the user to predict the likelihood of the infection.

How we built it

The microcontroller NodeMCU is programmed in C++ in Arduino IDE and gets the data from sensors through the I2C protocol. The Machine Learning model is trained on the Azure ML Studio, and the trained model is then deployed as a web service The dataset used was from this research paper - https://rdcu.be/cBdwN The Web app is built with flutter.

Challenges we ran into

The biggest challenge was to get the web app communicating with the microcontroller.

Accomplishments that we're proud of

The total cost of making the device prototype was around 10 USD even after buying everything from the retail if mass-produced the Per piece cost will come down even significantly.