Smart Wearables with Sensor Fusion for Fall Detection in Firefighting

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Emergency Response Systems

Background:

• Falls are a leading cause of firefighter fatalities in China, necessitating improved safety measures.
• Existing fall-detection systems (FDSs) often rely on single sensors and are not optimized for the unique challenges faced by firefighters, such as unstable environments and varied movements.
• Limited research addresses the impact of sensor placement on FDS performance, particularly for high-risk professions.

Purpose of the Study:

• To develop and evaluate a smart wearable fall-detection system (FDS) specifically designed for firefighters.
• To investigate the effectiveness of integrating multiple motion sensors (inertial measurement units - IMUs) at various body locations (chest, elbows, wrists, thighs, ankles).
• To assess the performance of a multisensory recurrent neural network in accurately detecting firefighter falls amidst complex movements and hazardous environments.

Main Methods:

• Integration of multiple inertial measurement units (IMUs) into firefighter personal protective clothing at five key body locations.
• Development of a multisensory recurrent neural network model for processing data from the integrated IMUs.
• Evaluation of different IMU combinations to determine optimal sensor placement for fall detection performance.

Main Results:

• The proposed multisensory recurrent neural network demonstrated high performance in detecting firefighter falls.
• Sensor fusion utilizing IMUs from all five proposed body parts achieved the highest performance metrics.
• The system achieved an accuracy of 94.10%, sensitivity of 92.25%, and specificity of 94.59% when using all five sensor locations.

Conclusions:

• A smart wearable FDS with strategically placed multiple IMUs offers a promising solution for enhancing firefighter safety.
• The multisensory approach, particularly with sensors on the chest, elbows, wrists, thighs, and ankles, significantly improves fall detection reliability.
• This technology can help mitigate risks associated with falls in challenging firefighting environments.