Parkinson’s Disease Induced Gain Freezing Detection using Gated Recurrent Units Optimized by Modified Crayfish Optimization Algorithm

Parkinson’s disease belongs to the group of health problems that are incurable but can be mitigated if treated properly. While there is no way of curing the damage caused by the disease, patient’s life quality can be improved if diagnosed and treated properly and on time. The role of artificial intelligence (AI) in medicine is increasing. Deep learning algorithms may be utilized to automatically detect freezing of gait episodes. This study is focused on Parkinson’s disease diagnosis based on the disturbances in the patient’s gait which is affected by this disease. A hybrid deep and machine learning AI solution that employs the gated recurrent unit (GRU) neural network optimized by a hybrid swarm intelligence solution between the crayfish optimization algorithm and the firefly algorithm has been proposed. The proposed solution is compared to other high-performing optimization algorithms to establish objective grounds for comparison. The proposed framework results in the overall best performance and confirms the made improvements. The best-constructed model attained an accuracy of 87.08%.