The Internet of Medical Things (IoMT) has emerged as a groundbreaking paradigm in healthcare, connecting billions of devices to enable enhanced automation, real-time communication, and comprehensive data collection. However, the rapid expansion of IoMT devices also presents significant security challenges that must be addressed through ongoing research and innovative approaches. In this study, we focus on improving attack classification in IoMT networks by leveraging federated learning, with a particular emphasis on the Random Forest (RF) model. Utilizing the open CICIoMT2024 dataset, which comprises 19 distinct attack classes and 45 features, we deploy the RF model across 10 simulated edge devices. The federated RF model demonstrated exceptional performance, achieving an accuracy of 0.9922, precision of 0.9938, recall of 0.9922, and an F1 score of 0.9909. This innovative approach significantly enhances security at the edge, particularly for devices with limited computational resources, underscoring the potential of federated learning to bolster IoMT security in increasingly complex healthcare environments.