Advances in interconnected healthcare devices—such as smart infusion pumps, wearable health monitors, and biosensors—are transforming patient care by enabling real-time vital sign monitoring. However, because these devices rely on wireless communication, they are highly susceptible to cyberattacks, including data interception and modification. Such threats not only put at risk the security of patient data but also pose significant risks to patient health. Our survey begins by examining the various types of attacks that target different layers of the IoT ecosystem, focusing on each layer’s vulnerabilities. Then, we show that other works have explored traditional intrusion detection techniques and centralized machine learning approaches, highlighting their limitations. Additionally, we present commonly used datasets for training and testing machine learning models. As its principal contribution, this survey provides a comprehensive overview of federated learning-based intrusion detection methods specifically used for medical IoT environments, offering insights into how this approach can strengthen security while safeguarding sensitive patient data.