Adverse drug events pose substantial challenges in healthcare, demanding effective prediction methods to booster patient safety. This paper introduces an innovative approach to adverse drug events prediction utilizing natural language processing techniques. Leveraging the WebMD Drug Reviews Dataset, our study employs a comprehensive methodology involving data acquisition, meticulous pre-processing, feature extraction, and classification with a Bayesian-Optimized Random Forest Classifier. This paper addresses limitations identified in extant literature, particularly concerning the accurate interpretation of intricate medical documents. Experimental results demonstrate promising performance, with a maximum accuracy of 77%, indicating the potential of our approach to enhance adverse drug events prediction and mitigate risks associated with medication use. This research aims to contribute to the advancement of pharmacovigilance practices and ultimately improve patient care standards.