"A Pocket-size electronic device with a Mobile app to display and analyze Body signals with a deep learning based interpretabile Multi-Model 🩺"
Project BioSense-AI is a both software - hardware project focusing on predicting diseases using ECG-PPG, temperature, and text prompts. Signals are amplified and filtered via custom analog circuits, then fed to Orange Pi Zero 2W SBC from an A2D conversion to digital processing. The system uses a deep learning custom multi-model architecture with interpretability for accuracy and efficiency. Currently, we are developing a mobile app for user access to the AI model by scanning ECG reports and other prompts for hospitals without our electronic device.
Specifications:
- Analog circuit design for capturing ECG signals.
- Rechargeable pocket size Electronic device.
- Connects to the Mobile app to display ECG signals and prediction reports.
- Classification and predictions of Lung Sounds (by stethoscope mic); Asthma, heart failure, pneumonia, bronchitis, pleural effusion, lung fibrosis, and chronic obstructive pulmonary disease (COPD)
- Classification and predictions of heart sounds (by stethoscope mic); Artefact, extra heart sound, extrasystole, murmur, and normal.
- Classification and predictions of ECG; -N: Non-exotic beats (normal beat) Supraventricular ectopic beats, Ventricular ectopic beats, Fusion Beats, Unknown Beats
Hardware Components:
- BioSense-AI main PCB
- Power Supply PCB
- Orange Pi 0 2w SBC
- Analog components
- USB, ECG, and Audio Ports
- OLED display for displaying essential states of the device
- Switches
Team Members:
Demonstration VideoLink: YouTube
Special thanks to Dr. Udaya Sampath Perera and Dr. Chamira Edussooriya for their guidance and support throughout our journey. Also, I would like to extend our special gratitude to Chamith Dilshan Ranathunga Aiya for the mentorship & invaluable support for the project. Department of Electronic and Telecommunication Engineering, University of Moratuwa