Parkinson's disease (PD) is incapacitating neurodegenerative disorder with a significant impact on individuals worldwide. Early detection of PD plays a vital role in providing timely treatment and improving patient outcomes. Recently, machine learning & deep learning techniques have shown promise in PD detection using various types of medical data, including MRI scan brain images and audio samples. In this work, a novel method for early detection of PD is developed. Aim of our work is to combine and use two different feature sets of input (MRI scan brain images and audio samples). Combining MRI and audio data holds promise as a valuable screening tool for early PD detection, enabling prompt treatment and improved patient outcomes. Both datasets are pre-processed, respectively, and given to the model for training. The Convolutional Neural Network (CNN) model is trained on MRI images and achieved an impressive accuracy of 98.0%, indicating its ability to correctly classify PD cases. On the other hand, the Cat Boost classifier algorithm yielded an even higher accuracy of 98.31% with audio data compared with XGBoost and Random Forest machine learning algorithms. Both Machine learning and Deep Learning models could potentially serve as valuable screening tools for the early detection of PD, allowing for timely treatment and better patient outcomes. The proposed model's ability to use both imaging and audio da- ta could enable a more efficient and accurate diagnosis of PD.