Parkinson’s disease (PD) affects global millions of people and early diagnosis is critical for effective treatment and improved quality of life. We propose a hybrid model that integrates XGBoost for feature selection and Long Short-Term Memory (LSTM) networks for time series analysis to identify PD using voice data. This system achieves high diagnostic accuracy validated through performance metrics and ROC-AUC. The hybrid approach outperforms standalone models and offers a scalable non-invasive and reliable diagnostic tool for clinical and telemedicine environments. By analyzing the voice data, we aim to provide a better understanding of the relationship between voice characteristics and Parkinson’s Disease, contributing to the development of reliable, non-invasive diagnostic tools. The integration of LSTM’s ability to handle sequential data with XGBoost’s efficient classification capabilities could offer a significant improvement in early detection methods, bringing us closer to accessible and accurate diagnostic solutions.