Advances in Detecting Parkinson's Disease

Diagnosing disordered subjects is of considerable importance in medical biometrics. In this study, aimed to provide medical decision boundaries for detecting Parkinson's disease (PD), we combine genetic programming and the expectation maximization algorithm (GP-EM) to create learning feature functions on the basis of ordinary feature data (features of voice). Via EM, the transformed data are modeled as a Gaussians mixture, so that the learning processes with GP are evolved to fit the data into the modular structure, thus enabling the efficient observation of class boundaries to separate healthy subjects from those with PD. The experimental results show that the proposed biometric detector is comparable to other medical decision algorithms existing in the literature and demonstrates the effectiveness and computational efficiency of the mechanism.