Deep learning framework for cardiorespiratory disease detection using smartphone IMU sensors

Area of Science:

• Biomedical Engineering
• Digital Health
• Cardiorespiratory Medicine

Background:

• Cardiovascular and respiratory diseases pose a significant global health challenge.
• There is a critical need for accessible, cost-effective screening tools for early detection.
• Current remote monitoring solutions often require specialized equipment or are not widely accessible.

Purpose of the Study:

• To develop and validate a smartphone-based framework for early detection of cardiorespiratory conditions.
• To leverage inertial measurement unit (IMU) sensors for non-invasive respiratory kinematics acquisition.
• To establish a cost-effective and accessible solution for remote health monitoring.

Main Methods:

• Utilized commodity smartphones with IMU sensors (accelerometer, gyroscope) for data collection.
• Implemented a standardized protocol involving data acquisition from five thoracoabdominal regions.
• Employed a bidirectional recurrent neural network (BRNN) for binary classification of healthy individuals versus patients with cardiovascular disease after signal preprocessing and breathing cycle segmentation.

Main Results:

• The BRNN model achieved robust classification performance: average sensitivity of 0.81±0.02, specificity of 0.82±0.05, F1 score of 0.81±0.02, and accuracy of 80.2%±3.9.
• The model demonstrated generalization capability on an independent dataset, achieving a true negative rate of 74.8%±4.5.
• The framework successfully differentiated between healthy individuals and preoperative patients with conditions like valvular insufficiency, coronary artery disease, and aortic aneurysm.

Conclusions:

• The proposed smartphone-based framework offers a promising, non-invasive, and low-cost approach for early cardiorespiratory condition detection.
• This technology is suitable for remote health monitoring, particularly in resource-limited settings and during public health crises.
• The study supports the potential for improving public health outcomes through enhanced early diagnosis and remote patient management.