The widespread uptake of the Internet of Medical Things (IoMT) has transformed healthcare by facilitating real-time monitoring and data-driven decision-making, but maintaining data privacy and security is a vital challenge because data breaches and unauthorized access are on the rise. Differential Privacy (DP) and Federated Meta-Learning (FML) are being seen as promising candidates to tackle these issues with the model performance maintained, wherein DP adds noise to sensitive data in a controlled manner for rigorous privacy assurance, and FML allows for personalized learning across distributed IoMT devices without the need for patient data centralization. This survey delves into the combination of DP and FML for preserving privacy in medical IoT use cases by presenting noteworthy methodologies like noise mechanisms, adaptive privacy budgets, and meta-learning strategies designed for diverse healthcare data. We also review state-of-the-art techniques, assessing their performance in maintaining privacy, avoiding adversarial threats, and maximizing model utility while presenting challenges like computational overhead, communication efficiency, and the privacy-accuracy trade-offs.