With the rising trend of polypharmacy in geriatric patients, the potential for adverse drug events increases, making the need for effective strategies in identifying high-risk prescriptions imperative. This study examines the performance of various machine learning models in predicting potentially inappropriate medications among older adults. Using electronic health records and pharmacy claims, this study addresses medication risk prediction by applying advanced computation techniques such as neural networks and random forests to analyze data and predict risks associated with the prescription of medications. It comprises data collection, preprocessing, model development, and evaluation with key performance metrics including accuracy, sensitivity, specificity, and the receiver operating characteristic curve’s area under the curve. Results showed that the neural network model outperformed the algorithms used by having an accuracy of 86.0% and an area of 0.90 under the curve, while the random forest model closely follows at a concerning accuracy of 85.5%. These results establish that it is one more domain in which machine learning turns out to be a disruptive technology for clinical decision support, enabling healthcare providers to identify at-risk prescriptions proactively for better medication safety among geriatric patients. The study ends with recommendations for incorporating computational models into clinical practice, attempting to improve prescribing practices and decrease adverse drug reactions among older adults. Directions for future research, therefore, focus on refining the performance of such a model towards better predictions and real-world outcomes in the treatment of patients.