Recall-driven Precision Refinement: Unveiling Accurate Fall Detection using LSTM

Rishabh Mondal,Prasun Ghosal

2023-09-10

Abstract:This paper presents an innovative approach to address the pressing concern of fall incidents among the elderly by developing an accurate fall detection system. Our proposed system combines state-of-the-art technologies, including accelerometer and gyroscope sensors, with deep learning models, specifically Long Short-Term Memory (LSTM) networks. Real-time execution capabilities are achieved through the integration of Raspberry Pi hardware. We introduce pruning techniques that strategically fine-tune the LSTM model's architecture and parameters to optimize the system's performance. We prioritize recall over precision, aiming to accurately identify falls and minimize false negatives for timely intervention. Extensive experimentation and meticulous evaluation demonstrate remarkable performance metrics, emphasizing a high recall rate while maintaining a specificity of 96\%. Our research culminates in a state-of-the-art fall detection system that promptly sends notifications, ensuring vulnerable individuals receive timely assistance and improve their overall well-being. Applying LSTM models and incorporating pruning techniques represent a significant advancement in fall detection technology, offering an effective and reliable fall prevention and intervention solution.

Signal Processing,Artificial Intelligence

What problem does this paper attempt to address?

The paper aims to address the issue of fall detection for the elderly by developing an accurate and practical fall detection system to tackle this global health challenge. Specifically, the research team proposed a wearable device that combines accelerometers, gyroscope sensors, and Long Short-Term Memory (LSTM) networks to achieve real-time detection of fall events.

To improve the system's performance, the researchers also introduced weight pruning techniques to optimize the architecture and parameters of the LSTM model, thereby reducing false alarms while ensuring a high recall rate. Additionally, the system integrates a Raspberry Pi-based hardware platform to achieve real-time execution capability and designed an intuitive Android application to assist caregivers in providing timely help.

Key contributions of the research include:
- Proposing a method that prioritizes recall rate over precision to ensure fall events are identified as accurately as possible and to reduce missed detections for timely intervention.
- Developing a novel fall detection system that combines the LSTM model and pruning techniques, significantly enhancing the accuracy and efficiency of fall detection technology.
- Implementing an integrated wearable node that combines fall detection, victim localization, and staff notification functions.
- Designing an easy-to-use Android application to help caregivers provide necessary care and support.

Experimental results show that the system has a high recall rate (95%) and good specificity (96%), and it performs excellently in Receiver Operating Characteristic (ROC) curve analysis, with an Area Under the Curve (AUC) of 0.96, demonstrating the system's effectiveness in distinguishing between fall and non-fall events.

In summary, this research provides an advanced solution for elderly fall detection and is expected to improve the quality of life for the elderly by reducing fall risks. Future research directions will include further model optimization, integration of more sensors, and expansion of the dataset.