Doppler Radar Sensor-Based Fall Detection Using a Convolutional Bidirectional Long Short-Term Memory Model.

Area of Science:

• Gerontology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Falls among the elderly represent a significant health risk, necessitating advanced detection methods.
• Radar-based systems offer a promising non-intrusive approach for continuous elderly fall monitoring.

Purpose of the Study:

• To develop and validate a novel deep learning model for accurate radar-based fall detection in the elderly.
• To enhance the reliability and accuracy of fall detection systems by analyzing radar signal frequency spectrums.

Main Methods:

• A convolutional bidirectional long short-term memory (CB-LSTM) deep learning model was designed.
• The model processes the frequency spectrum of radar signals to capture temporal and spatial features.
• Extensive experiments were conducted for performance evaluation and comparison.

Main Results:

• The proposed CB-LSTM model achieved a fall detection accuracy of 98.83%.
• The model demonstrated superior performance compared to existing relevant fall detection methods.
• The system effectively utilizes radar frequency spectrums and deep learning for fall monitoring.

Conclusions:

• The developed radar-based fall detection system provides effective technical support for monitoring elderly falls.
• This technology has significant potential to improve the quality of life for seniors and enable timely rescue.
• The CB-LSTM model offers a reliable and accurate solution for elderly fall detection.