Advancements in medical foundation models are revolutionizing healthcare by enabling more personalized and interpretable patient care. We introduce MambaHealth, a lightweight and efficient foundation model designed to address complex healthcare challenges through task-specific adaptation. MambaHealth’s pretraining integrates diagnostic and procedural data to derive detailed patient state representations, which are then fine-tuned for applications such as drug recommendation, multi-diagnosis management, and temporal prescription optimization. A key feature of MambaHealth is its emphasis on explainability, providing transparent and interpretable insights into its decision-making processes, thereby enhancing trust and reliability in clinical environments. Moreover, MambaHealth offers personalized recommendations based on individual patient data, ensuring adaptability to each patient’s unique characteristics. By continuously refining its parameters with updated clinical data, MambaHealth consistently outperforms existing models in both accuracy and efficiency, making it a valuable tool for advancing intelligent healthcare management and supporting informed clinical decision-making.