Advancing Human Fall Detection Through Motion Data from Smart Wearable Sensors and CNN-BiGRU Model

Falls pose a significant health risk, especially for older adults, resulting in serious injuries, limited movement, and higher medical expenses. Precise and fast identification of falls is essential for immediate healthcare treatment and avoidance of additional complications. This paper presents a new method to improve the identification of individual falls by using movement inputs from smart wearable sensors and a hybrid deep learning model that combines convolutional neural networks (CNN) and bidirectional gated recurrent units (BiGRU). Intelligent wearable sensors, including accelerometers and gyroscopes, are used to collect real-time movement data from different parts of the human body. The movement signals gathered are subjected to preprocessing and segmentation into windows of a particular length to extract pertinent information. An advanced neural network structure comprising CNN layers for extracting spatial features and BiGRU layers for representing temporal sequences is specifically intended to acquire the distinctive patterns related to fall occurrences efficiently. The proposed CNN-BiGRU model is trained on a varied dataset that includes simulated falls and activities of daily living conducted by several people. The model's effectiveness is assessed using established measures, such as accuracy, sensitivity, specificity, and F1-score. The experimental findings demonstrate that the CNN-BiGRU model has a notable level of accuracy in recognizing instances of falls. Combining in-telligent wearable sensors and advanced deep-learning algorithms creates a dependable and highly effective system for detecting falls. The suggested strategy can significantly enhance the well-being of those prone to falls by offering prompt aid and mitigating the impact of injuries caused by falls.