Fog-Cloud Enabled Human Falls Prediction System Using a Hybrid Feature Selection Approach

Elderly people's human fall prediction is identified as one of the more challenging factors in real-time healthcare monitoring systems. This type of healthcare system creates huge traffic and delays due to continuous data transmission from the sensing device to a cloud-based processing system. So, a novel fog-cloud-enabled human fall prediction system is proposed to minimize traffic and quick response time due to closer decision-making in the fog-level computing environment. Then, the amount of data sensed through the accelerometer sensors deployed over humans can be transferred from fog to the cloud computing layer. To minimize the data transfer from fog to cloud layer, the fall prediction system incorporates a hybrid feature selection approach using Particle Swarm Optimization and Grey Wolf Optimization (PSO-GWO). This novel feature can significantly minimize the bandwidth usage and latency between the fog and cloud computing nodes. As a result, the proposed fall prediction system significantly improves prediction time and accuracy compared to the existing fall prediction systems.