With the rapid development of medical information technology, the amount of medical data continues to increase, and the data structure becomes increasingly complex. How to efficiently process and utilize this data to improve the quality and efficiency of medical services has become an important issue. This article proposes an innovative design for a medical big data platform that integrates machine learning and knowledge graph, using large-scale language models and deep learning models to conduct deep analysis and mining of medical text, images, and other data; Adopting a knowledge graph based medical data integration method to build a sustainable medical big data ecosystem. By transforming medical data from different sources and categories into a unified knowledge representation, the integration, storage, management, analysis, and mining of medical data can be achieved. The research results will provide more accurate, faster, and effective data decision-making support for applications such as hospital management, clinical treatment, and scientific research and teaching.