Early identification of health irregularities is critical for preventing serious medical conditions, but continuous monitoring of vital parameters such as heart rate and blood oxygen levels is difficult without wearable, automated systems. Traditional health checks rely on periodic measurements and manual observation, which can miss subtle or early-stage anomalies. Real-time detection also demands fast processing and minimal latency, which cloud-dependent systems may not always provide.
This project presents a compact, edge-based AI health monitoring system that performs real-time anomaly detection directly on the device. Using an on-device machine learning model trained via Edge Impulse, the system learns normal health patterns and identifies deviations instantly. The IndusBoard Coin handles sensor data acquisition, local ML inference, and display output, enabling continuous monitoring without reliance on external computing resources.
Key Features
- Edge ML-based anomaly detection with real-time inference
- MAX30100 sensor for heart rate and SpO₂ measurement
- On-device processing without cloud dependency during operation
- Compact 3cm IndusBoard Coin with onboard battery connector
- Built-in sensors for temperature, light, and magnetic field sensing
- OLED display support for real-time vitals visualization
- ML model training and deployment using Edge Impulse
Applications
- Wearable health monitoring and personal wellness tracking
- Early detection of irregular heart rate or oxygen saturation patterns
- Research and prototyping for medical IoT devices
- Continuous vitals monitoring in remote or low-connectivity environments
- Educational projects involving edge AI and biomedical sensors
This implementation highlights how edge machine learning and compact embedded platforms can be combined to enable continuous health monitoring and anomaly detection. Boards such as the IndusBoard Coin, with integrated sensing, battery support, and sufficient computational capability, are well suited for developing and testing wearable health applications. Projects like this provide a practical framework for exploring real-time AI inference, sensor data analysis, and low-power system design in healthcare-focused embedded systems.