news

"ai is becoming an important force in promoting the innovation and development of my country's medical system. as an important tool for promoting new informatization and cultivating new quality productivity, ai has also become a breakthrough in solving the difficulties in the medical field. 'ai+medical' is expected to become the next major economic growth point in the future." hu zhenning, director and deputy general manager of runda medical, said in a recent interview with a securities times reporter.

ai is penetrating into various industries at an unprecedented speed and depth, and the medical field is one of the areas where ai has exerted its transformative power most significantly. currently, ai is being widely used in various medical fields, including imaging diagnosis, auxiliary clinical decision-making, health management, etc., and has significant advantages in improving medical quality and efficiency, personalized treatment, data processing, etc.

with the development of social economy, the intensification of population aging and the improvement of residents' health awareness, residents' demand for high-quality medical resources has also risen rapidly. however, under the traditional medical service model and medical resources, my country has always faced the dilemma of relatively insufficient total medical and health resources and uneven distribution. "ai+medical" is the "solution" to help solve this problem.

hu zhenning pointed out that the widespread application of ai in the field of medical diagnosis and treatment is an inevitable trend. on the one hand, it is based on the background of tight supply of medical resources in hospitals, and on the other hand, it is due to the demand of residents for high-quality medical resources. multiple factors determine that the ai ​​medical field has a large market capacity. "ai+medical" is breaking the time and space barriers of technology and region in the traditional medical field, allowing high-quality medical resources to transcend geographical boundaries.

in recent years, the development momentum of digital medical and health technology in my country has been relatively rapid, and the application scenarios and market size are increasing rapidly. according to the global market insights report, the "ai+medical" market size is expected to grow at an average annual compound growth rate of more than 29%, reaching us$70 billion in 2032, equivalent to rmb 500 billion.

the most important point of ai empowering medical diagnosis and treatment is that it can reduce costs, which is expected to solve the common pain points of residents, hospitals, and governments. for example, for medical institutions, the application of ai big models can help improve management efficiency, reduce conflicts between doctors and patients, and help control medical insurance costs.

hu zhenning said, "with the help of the application of ai big models, the problem of 'waiting for three hours and seeing a doctor for three minutes' that residents often encounter is expected to be solved, and the health literacy of residents can also be effectively improved; for hospitals, it takes decades to train a senior doctor, and the speed of training doctors is far slower than the speed of residents' increasing demand for high-quality medical resources. the application of ai in hospital diagnosis and treatment is equivalent to the doctor deriving a digital avatar, which comprehensively solves the problems of residents' difficulty in seeing a doctor and the tight supply of medical resources in hospitals."

recently, runda medical released the medical big model full-scenario application "cdx good doctor xiaohui" equipped with huawei cloud big model, which mainly empowers all medical application scenarios including clinical, patient services, scientific research, etc., and provides comprehensive services for medical staff and individuals.

after the release of the above-mentioned medical big model, hu zhenning admitted that in the past decade of exploration, runda medical has formed two main cognitions, which are also the two major difficulties that the medical big model must overcome. that is, medical data are mostly natural language texts, which are difficult to be effectively processed by machines and cannot reflect the value of data elements; in addition, in medical scenarios that require evidence-based treatment, how to suppress the inherent illusions of big models.

"there are various bottlenecks in the application of ai in the field of diagnosis and treatment. it is not logically difficult to train a large ai model into a doctor's brain, but it requires solid basic data. if there is a lack of basic data, then this large model can only provide a diagnosis and treatment plan that is not accurate enough and has no specific value." hu zhenning said.

he pointed out that data is the foundation of artificial intelligence applications. medical data is mainly in hospitals, and most hospitals are public hospitals. for the government and hospitals, how to activate huge data assets is a key step in studying the implementation of ai in the medical field.

"the decade-long experience in data governance is a major advantage for runda medical in developing medical big models, which also lays a solid foundation for the creation of ai big models." according to hu zhenning, in response to the technical difficulties in medical data governance, runda medical has created cdx (clinical data x, also known as the "medical data base") based on the nearly 100,000 data annotations and governance rules accumulated in the past six years, through training the huawei cloud pangu big model. this allows the medical community to no longer be restricted by data field design based on expert experience, and all clinical text records can be converted into "fully structured data."

in terms of overcoming the illusion of big models and providing evidence-based basis, runda medical has built the l1 medical vertical big model based on the l0 pangu general big model, and formed a medical team, hiring clinicians, experts and consultants from dozens of well-known hospitals such as beijing fuwai, huashan, shengjing, tongji union, ruijin, changzheng and changhai to participate in the polishing of big models and application products.

focusing on the overall application of the medical diagnosis and treatment industry, hu zhenning believes that it is not difficult to widely apply big models in the industry, but the difficulty lies in how to integrate big models with the industry more deeply. "nowadays, if you ask some general questions about symptoms, the big model can answer them, but for very specific questions, the big model is currently at a stage where it cannot provide diagnosis and treatment plans because it does not have enough knowledge. if the big model is to be deeply applied in the medical industry, it still needs people to continue to refine it, and the best way is to cooperate deeply with hospitals and the government."

the rapid development of ai in the medical field means that china's medical services will be further accelerated in terms of intelligentization and digital transformation. "the application of ai big models in the medical field is a bit like navigation at present, but i believe that one day it will grow into driverless driving, which is also what the industry expects of it," said hu zhenning.