<p>The growing intricacy of healthcare processes, intertwined with the rising need for personalized therapies, necessitates innovative technological solutions. Artificial Intelligence (AI) has displayed an outstanding potential in healthcare through optimizing therapeutic interventions, aiding in medical decision-making, and promoting patient monitoring. However, the integration of AI in healthcare faces critical challenges, including data interoperability, security concerns and the need for systems that can support real-time and large-scale deployments. A promising architectural solution to these challenges lies in the adoption of microservices. Existing healthcare recommendation systems often suffer from interoperability issues due to diverse healthcare standards, multiple protocols, and systems incompatibilities. Furthermore, ensuring secure data exchange while maintaining efficiency remains a key hurdle. To address these challenges, we set forward PAM-SHRM (Pretrained AI Model Based on Microservices for Secure Healthcare Recommendation Systems), which leverages a microservice architecture to boost scalability, modularity and security in AI-driven healthcare solutions. Our contribution lies in handling interoperability challenges by structuring AI-based recommendation engines within a microservices architecture. It enhances security and privacy through robust data protection measures integrated within the architecture. It fosters healthcare recommendation accuracy by using pre-trained AI models optimized for personalized medicine. In addition, the model was trained and evaluated on BioASQ, a diverse and representative healthcare dataset, which further strengthens its ability to generalize across real-world medical scenarios and supports the personalization of recommendations.</p>