Currently, artificial intelligence (AI) technology has shown great potential and value in fields such as disease diagnosis, image recognition, and drug development. How to further promote the deep empowerment of medical development by artificial intelligence, and how to make artificial intelligence play to its strengths and avoid its weaknesses in empowering medical development? On April 19-20, the 2025 China Medical Development Conference with the theme of "Empowering Medical Development with Artificial Intelligence" was held in Beijing. The attending experts discussed innovative paths and development directions for the deep integration of artificial intelligence and medicine. Wang Chen, Vice President of the Chinese Academy of Engineering and President of the Chinese Academy of Medical Sciences, pointed out that artificial intelligence is profoundly changing the paradigm and industry ecology of medical research. The medical community needs to adapt to the development of artificial intelligence, promote the integration of artificial intelligence and medicine in stages, and leverage the important role of artificial intelligence as a new quality productivity to care for people's health. The medical field data has characteristics such as sensitivity, irreversible results, and complex responsible parties Wang Jiangping, former deputy minister of the Ministry of Industry and Information Technology, frankly stated that the particularity of the medical and health industry requires the comprehensive penetration of human-machine alignment rules in the medical and health AI model, that is, through technological means and ethical frameworks, to ensure that the target behavior and output of AI are consistent with human values and social norms. Starting from algorithms and data, we need to build a good "underlying architecture". "With the development of artificial intelligence and the emergence of large amounts of data, the conditions for studying biology and medicine through mathematics have become mature," explained Qiu Chengtong, a professor and mathematician at Tsinghua University. Medical data has nonlinear characteristics, and traditional artificial intelligence models are usually based on linear assumptions, making it difficult to capture the complex nonlinear relationships in medical data. Faced with the challenge of medical data, mathematics, medicine, artificial intelligence and other fields should strengthen cooperation, provide underlying architecture for artificial intelligence, and continue to promote the application of artificial intelligence in multiple fields such as disease research. With the support and improvement of mathematical science, some medical events have been better analyzed. For example, for the typical nonlinear event of "aging" in life activities, a study successfully presented its occurrence and development patterns after analyzing over 240 billion data points using multiple mathematical analysis algorithms. Data issues remain a bottleneck that hinders the deep application of artificial intelligence in the medical field. Qiu Chengtong said that medical data has problems such as "noise" (incorrect or irrelevant information in the dataset) and missing values. Traditional methods often require complex preprocessing when dealing with these problems, which increases the complexity and uncertainty of data processing and affects the accuracy of the model. Shen Jianfeng, a first level researcher at the Planning, Development, and Information Technology Department of the National Health Commission, believes that it is necessary to build high-quality datasets and artificial intelligence corpora in the medical field to break through technical bottlenecks such as insufficient professional corpora and inconsistent multimodal processing. The relevant departments are strengthening the mining of data value, improving diagnosis and treatment efficiency and accuracy through measures such as standardization construction, cross institutional data sharing, and vertical model application development. Facing safety and governance, building a new medical ecosystem for safety and governance is an important issue in the development of medical artificial intelligence. Physiological data, medical history information, and other medical data are crucial to the privacy and right to life of every patient. How can we avoid giving up on eating due to choking and make full use of medical data while protecting safety? We should establish data filters to automatically block data sources that violate ethics, and build a trustworthy medical data space to promote data sharing and circulation Wang Jiangping emphasized that privacy protection must be placed in a prominent position in dataset construction, deeply integrating medical expertise with advanced data science and technology, and effectively addressing issues such as sensitivity and fragmentation of medical data through privacy computing, terminology standardization, multimodal association, and small sample augmentation strategies. Zhao Wei, Director of the Statistical Information Center of the National Health Commission, emphasized that the application of medical artificial intelligence should always be based on safety and trustworthiness, adhere to the principle of patient-centered and doctor centered medical decision-making, establish a dynamic and sustainable evaluation mechanism, and promote the coordinated governance of law, technology, and ethics. To enable AI to make a leap from efficient tools to trusted partners, medical AI applications need to achieve alignment goals of interpretability, trustworthiness, and human harmony Wang Jiangping said that the alignment rule should be deeply integrated into the technical architecture, dataset construction, hospital management, patient awareness, industry supervision, and other aspects of medical AI. When it comes to the development path of empowering medicine with artificial intelligence, Wang Chen suggests that the task should be clearly defined and promoted in stages. Clarify the positioning and application logic of artificial intelligence in medicine in the short term, and carry out scenario based pilot projects; Expand the application scope in the mid-term stage and promote the full scenario integration of artificial intelligence and healthcare; Long term development should cover the entire chain of scientific research, clinical practice, and management, build a new ecology empowered by artificial intelligence in medicine, and participate in international rule making. (New Society)