The probablistic random forest clinico-statistical regression analysis of MER signals with STN-DBS and enhancement of UPDRS

Abstract

In this study, we present classification and regression analysis to predict the UPDRS score and its enhancement after the microelectrode STN signal recording (MER) with DBS surgery (implantation of the microelectrode). We hypothesized that a data informed grouping of features extrapolated from MER signals of STN can envisage restore (by decreasing the tremor) and functioning the motor improvement in Parkinson’s disease (PD) patients. A random—forest is used to account for unbalanced datasets and multiple observations per PD subject, and showed that only five features of STN-MER signals are sufficient and account for prognosting UPDRS advancement. This finding suggests that STN signal characteristics are maximum correlated to the extent of improvement motor restoration and motor behavior observed in STN DBS.