Deep Learning enabled Fall Detection exploiting Gait Analysis.

Area of Science:

• Gerontology
• Computer Science
• Biomedical Engineering

Background:

• Falls are a major global public health issue, particularly with an aging population, leading to increased healthcare costs, reduced mobility, and potential fatalities.
• Existing fall detection methods may lack accuracy or efficiency in real-world scenarios.

Purpose of the Study:

• To develop and evaluate a novel Deep Learning enabled Fall Detection (DLFD) system leveraging gait analysis for improved fall detection accuracy.
• To provide a robust framework for real-time fall detection that can alert emergency services promptly.

Main Methods:

• A DLFD framework was proposed, extracting gait features from RGB videos using the MediaPipe framework.
• A normalization algorithm was applied, followed by classification using a bi-directional Long Short-Term Memory (bi-LSTM) model.
• The model was trained and tested on three public datasets comprising over 1 million frames of various activities and fall types.

Main Results:

• The DLFD model achieved a high accuracy of 96.35% in detecting falls across diverse datasets.
• The experimental results demonstrate the effectiveness and robustness of the proposed gait analysis-based fall detection method.

Conclusions:

• The developed DLFD system shows significant potential in mitigating the adverse effects of falls in the elderly population.
• Immediate alerting capabilities can expedite assistance, reduce prolonged injury, and ultimately save lives.