Wearable Breathing Sensor Uses AI to Classify Respiratory Patterns with High Accuracy
A new study published in Scientific Reports details the development of an advanced wearable breathing sensor that leverages artificial intelligence to monitor respiratory health.
How the System Works
The sensor combines an inertial measurement unit (IMU) with a flexible resistive sensor into a single wearable patch. This patch, attached to the chest, transmits data wirelessly via Bluetooth Low Energy to a processing unit.
The system uses a transformer-based deep learning architecture to classify respiratory patterns, achieving a validation accuracy of 93.41% for three-class classification and 78.57% for six-class classification on holdout data.
Study Results
The research team tested the system on 20 healthy adult participants. When comparing its performance to other machine learning models, the transformer model outperformed both CNN-LSTM and histogram gradient boosting classifiers.
Potential Applications
The study highlights the system’s potential for remote monitoring of chronic respiratory conditions, including:
- Sleep apnea
- Asthma
- Chronic bronchitis
Key Takeaway
The wearable patch successfully classifies complex respiratory patterns with high accuracy, demonstrating a clear advantage over traditional deep learning and machine learning approaches. This points toward a future where continuous, passive monitoring of breathing health could be performed remotely without cumbersome equipment.