parkinson’s disease is a progressive neurodegenerative disorder that results in the gradual degeneration of brain cells and significant impairment in motor functions. Early and accurate diagnosis of Parkinson’s disease is crucial for timely medical intervention and improved quality of life. In recent years, the use of intelligent computational techniques has gained increasing attention for the early detection of this disease. Among the widely adopted algorithms in this domain are Support Vector Machines (SVM), k-Nearest Neighbors (KNN), and various types of Neural Networks. Despite their success, achieving a robust and optimized combination of these algorithms remains a challenging task.

In this research, a novel hybrid model is proposed that integrates an optimized Support Vector Machine with a Deep Belief Network (DBN). Given the sensitivity of SVM performance to parameters such as the kernel width (sigma) and penalty parameter (C), Particle Swarm Optimization (PSO) was employed to optimize these parameters. The output of the optimized SVM was then used as the input for the DBN, enabling the model to learn more complex patterns. Experimental evaluations on benchmark datasets demonstrated that the proposed hybrid method (PSO-SVM + DBN) outperformed standalone models and their non-optimized combinations. The model achieved a test accuracy exceeding 99%, indicating its high potential and effectiveness for early-stage Parkinson’s disease diagnosis.