Scopus Research — abdulrahim thiab humod ithawi

Falls that occur among elderly adults or individuals with mobility challenges may provide certain health hazards. Due to this, effective fall detection systems need to be designed that are accurate and real-time. The classic methods of fall detection are threshold-based methods and rule-based methods that are less suited to diverse environments and generate too many false alarms. To improve on these disadvantages, this research develops a deep learning-based hybrid model employing CNN and LSTM networks to obtain increased fall detection performance. This model integrates both vision-based data and sensor-based data; it learns on benchmarks. This boosts the detection rate. The experiments prove that the suggested CNN-LSTM model is superior in terms of accuracy, precision, recall, and F1-score. The model's accuracy reaches 96.4 % and also has substantially lower false positive and false negative rates than state-of-the-art ones. The comparison with 3D-CNN, Bi-LSTM, and Hybrid DNN models indicated that our system excels in real-time latency and robustness in various situations. The proposed system can be used for geriatric monitoring, smart health care systems, and autonomous emergency response. However, the study has several drawbacks, such as being computationally expensive and requiring big and diverse datasets to generalize properly. Future studies will work on merging several types of sensors, learning capabilities, and edge-based computers so as to shorten real-time detection. The outcomes of this research can help in the design of AI-based medical instruments that are safer for sensitive parts. © 2025 IEEE.