Hackers are employing increasingly advanced methods to target hospital networks, electronic health data, medical IoT devices, and telemedicine apps as the healthcare industry becomes more digitally integrated. Gathering and sharing private medical information is essential when employing centralized machine learning to identify intrusions. Important concerns are privacy, regulations, and compliance with HIPAA, GDPR, and other laws. Federated Learning (FL) is an approach to learning that protects patient data and does not depend on a central server. It makes it possible for different healthcare facilities to work together to create efficient security models. This study looks at the ways that FL can enhance network security and preserve privacy in the healthcare industry. It explores secure aggregation techniques, federated learning systems, differential privacy strategies, and strong defences against inference and poisoning attacks. By using a hierarchical federated framework, this study shows how multi-provider healthcare organizations can increase detection accuracy while preserving their independence, privacy, and compliance.