This is a work-in-progress project. I will be updating the work as soon as progress is made.
Project end goal:Analyze sleep quality by continuously monitoring heart rate, oxygen saturation, movements, and snoring patterns and make this data available in a Bluetooth mobile application.
Stages:
- Capture live audio on the nrf5340-DK using a MEMS microphone (SPW2430), and classify it as snoring or no-snoring on the board
- Capture heart rate and oxygen saturation using a pulse oximetry and heart- rate monitor sensor (MAX30100)
- Capture body movements using an accelerometer sensor (ADX345)
- Develop a mobile app to display the above data send from the nrf5340-DK via Bluetooth
- Platform used for training and neural network model conversion: Edge Impulse
- Neural Network used: Depthwise separable convolution based network
- Dataset used for training: Snoring data from AudioSet project of google (https://research.google.com/audioset/dataset/snoring.html)
- Adapted the edge-impulse wrapper code provided by Edge impulse to run the neural network on the nrf5340-DK (https://docs.edgeimpulse.com/docs/running-your-impulse-locally-zephyr)
- Developed using MIT App Inventor
- Adapted the peripheral_hr sample project provided by zephyr to send data to Bluetooth app. (https://developer.nordicsemi.com/nRF_Connect_SDK/doc/latest/zephyr/samples/bluetooth/peripheral_hr/README.html#peripheral-hr)
Hardware interfacing (MAX30100, SPW2430, ADX345) is in progress
Analysis of power used and further optimizations need to be done after the completion of hardware interfacing.
Comments
Please log in or sign up to comment.