Since 2016, when Google's AlphaGo beat Go world champion Lee Sedol, the popularity of ARTIFICIAL intelligence has been high. Ai is not just playing chess, it has gradually penetrated every aspect of our lives. As the most popular, the industry has high expectations and relatively mature field - medical artificial intelligence, is considered to be the most promising application.
I. National policies on ARTIFICIAL intelligence
Since 2015, China has successively issued a series of policies to promote the development of medical ARTIFICIAL intelligence, which has played a guiding role in the application and development of artificial intelligence in the medical field. The application of artificial intelligence in medical treatment is mainly divided into the following directions: clinical diagnosis and treatment decision support system, intelligent medical image recognition, pathological classification and multidisciplinary consultation, intelligent voice electronic medical record system, etc.
In April 2017, the National Health and Family Planning Commission (NHFPC) issued the Guiding Opinions of The General Office of the State Council on Promoting the Construction and Development of The Medical Consortium, which pointed out in accordance with the spirit of the relevant documents of comprehensively implementing the hierarchical medical system in the 13th Five-Year Plan: Building clinical decision support system with artificial intelligence technology and sinking standardized treatment to the grassroots is one of the effective ways to solve the current shortage and unreasonable allocation of medical resources and the problem of people's difficulty in seeing a doctor, which is also in line with the good vision of healthy China. The following is an in-depth discussion on the application status and future development trend of artificial intelligence clinical decision assistance system.
Second, the domestic application status of ARTIFICIAL intelligence CDSS
Clinical Decision Support System (CDSS) is a medical information technology application System based on human-computer interaction. It aims to provide Clinical Decision Support (CDS) for doctors and other health practitioners, and complete Clinical Decision through data and models.
CDSS originated in the United States, with an estimated market size of nearly $500 million in 2018, the application of CDSS can reduce the probability of medical errors caused by improper medication or improper operation and reduce unnecessary harm to patients. CDSS is an important means to improve medical quality. Its fundamental purpose is to evaluate and improve medical quality, reduce medical errors and control medical expenses.
According to the system structure, CDSS can be divided into two types: knowledge-based CDSS and non-knowledge-based CDSS.
CDSS based on knowledge base generally consists of three parts: knowledge base, inference machine and man-machine interface. The knowledge base stores a large amount of compilation information, the inference machine integrates and analyzes the data automatically according to the rules of the knowledge base, and the man-machine communication interface feeds back the analysis results to the user, and can also serve as the system input, which is mainly used to meet the user's query requirements. This type of CDSS is relatively closed and lacks machine deep learning function, so all information collection, compilation, collation and rules need to be completed manually, resulting in high maintenance costs and poor timeliness of information update.
CDSS based on non-knowledge base generally adopts the form of artificial intelligence, which relies on artificial neural network and has the ability of machine learning. It can summarize and clarify knowledge in the process of human-computer interaction and continuous training, and provide suggestions for users with knowledge. With the gradual improvement of science and technology and informatization of the medical industry, the connection of electronic medical record system -CDSS- Internet database can be used to access tens of thousands of documents in an instant. This type of CDSS is bound to become the development trend in the future by providing accurate decision-making suggestions through efficient learning ability. The development of intelligent decision making systems will help clinicians keep abreast of medical developments, grasp evidence from evidence-based medicine, and respond more fully and freely to clinical problems.
As far as the application of domestic institutions and clinicians are concerned, KNOWLEDGE-based CDSS is still the mainstream. However, with the development of artificial intelligence and other relevant computer technologies, training machines to replace human beings in repetitive labor, CDSS based on non-knowledge base is the trend of future development. Domestic medical technology enterprises also focus on CDSS based on non-knowledge base. However, because CDSS based on non-knowledge base needs continuous training, this process is still quite labor-intensive and time-consuming according to the current technical means. It usually takes several years to train a disease, and the current products on the market have not entered the stage of mature application, which still has certain limitations:
Although after several years of development, learning and training, currently the supported diseases are still limited, and the scope of support for single diseases is also limited, and the support degree for complex diseases or diseases is not good.
The training methods and training logic have a great influence on the decision results. Medicine, especially oncology, is a very complex discipline, and the treatment methods of many problems are still controversial in today's academic circles, resulting in the situation that oncologists do not understand or recognize some decision results.
Due to different policies, medical insurance coverage and treatment costs, the use and recommendation order of drugs are also different in different countries and regions. If the current standard of only one country is applied, it will be difficult to meet the application needs of the whole world.
The development trend of CDSS in China
The existing difficulties and future development direction of CDSS need to be expanded from three dimensions.
Technical dimension. To strengthen the construction of artificial intelligence neural network, improve its learning, retrieval, calculation, analysis and other capabilities, and transform from training-dependent to self-learning, will better represent the opinions of artificial intelligence rather than trainers.
The medical dimension. Further expand the coverage of products, deepen the scope of application of products and expand the scope of database search, and at the same time carry out more targeted medical logical thinking training and conditional weight analysis training, so that artificial intelligence can truly understand medical records and medical literature, rather than relying on simple judgment of conditions to get results.
Product dimension. More consideration should be given to the application scope and application scenarios of products, and user experience should be improved from the perspective of users, and users' real needs and pain points should be integrated into products.
Based on the combination of big data, the functions of CDSS can be expanded to a broader space in the future, such as hospital/department management, scientific research collaboration platform building, structured medical record system, patient interaction and patient education, doctor continuing education, pharmacovigilance, medical cost control and other directions. On the basis of breaking through technical barriers, standardized treatment is promoted from top to bottom in accordance with clinical application scenarios, so as to improve the quality and efficiency of medical services and promote the healthy development of the medical industry ecosystem.
As the second largest medical market in the world, China has a huge market potential for CDSS. It is believed that with the strong support of the government and the continuous innovation of the technology industry, the real medical artificial intelligence will no longer be a plot in science fiction movies, but a foreseeable future in the short term. CDSS is expected to enter the hospital to assist doctors, save the labor force of doctors, and bring better medical experience to patients. It is really in line with the principle that "minor diseases do not go out of the township, major diseases do not go out of the county, and it is very convenient to see a doctor", so as to realize the real hierarchical diagnosis and treatment.
