Xinhua News Agency, Guangzhou, February 19 (Reporter Xiao Sisi, Zhou Lin) can "read the map" to identify images, but also "recognize the word" to read the medical records, and even "think" like a doctor, issue a diagnosis report, give treatment advice... ...This is not science fiction, artificial intelligence medical is transforming from cutting-edge technology to real-life applications, and "artificial intelligence doctors" are getting closer and closer to patients.

　　The agency predicts that the market demand for medical artificial intelligence in China has reached tens of billions of yuan.Experts believe that the application of "artificial intelligence doctors" is conducive to alleviating the imbalance between supply and demand of medical resources and uneven geographical distribution caused by the aging of society. So, when will the “Artificial Intelligence Doctor” actually on duty?

　　Artificial intelligence + medical market size continues to grow

　　The team led by experts such as Xia Huimin, a professor at the Guangzhou Women and Children Medical Center, and Zhang Kang, a professor at the University of California, San Diego, developed the “Auxiliary Bear” artificial intelligence diagnosis platform to automatically learn 1.36 million high-quality electronic texts of 567,000 children. The diagnostic logic in the medical record diagnoses a variety of common pediatric diseases with an accuracy comparable to that of an experienced pediatrician.On February 12th, the results of this clinical intelligent diagnosis research were published online in the well-known medical research journal Nature Medicine.

　　Dr. Yang Zhiming, CEO of Beijing Deep Thinking Artificial Intelligence, believes that artificial intelligence technology is constantly evolving and is being combined with medical vertical application scenarios.

　　The combination of artificial intelligence and medical care is seen as one of the investment hotspots of the next 5-10 years.According to the report of the Prospective Industry Research Institute, the scale of China's artificial intelligence + medical market size continues to grow. In 2017, it exceeded 13 billion yuan, an increase of 40.7%. In 2018, the market size was about 20 billion yuan.So what can an “artificial intelligence doctor” do?

- Alleviate the strain on medical human resources. In the Department of Radiology of the Ninth People's Hospital of Shanghai, only the lung examinations can reach 150 people per day.After the hospital introduced the lung cancer image intelligent diagnosis system, this artificial intelligence technology compressed the lung image diagnosis to the second level.Xu Ruihua, dean of the Cancer Center of Sun Yat-Sen University, believes that artificial intelligence can alleviate the shortage of doctor resources to a certain extent, so that high-quality medical resources can sink to the grassroots level, so that more people can enjoy Inclusive medical care.

　　- Prevent chronic diseases. Due to the low accuracy of chronic disease screening, the difficulty of targeted interventions, and the lack of health management tools, the awareness rate of diabetes in adults is only 30.1%.Shanghai Ruijin Hospital and Artificial Intelligence Company's fourth paradigm jointly launched a diabetes-based and complication management product based on artificial intelligence. Entering personal information in the Subscription, including gender, weight, fasting blood glucose, etc., can predict the risk factor for diabetes in the 3 years of this person.

　　- Improve the efficiency of cancer screening. Early diagnosis and early treatment is the key to improving the cure rate of cancer.In December 2018, the Cancer Center of Sun Yat-sen University led the development and application of the artificial intelligence diagnosis and treatment system for upper gastrointestinal cancer. According to the initial data analysis of the system, the accuracy of malignant tumor recognition in clinical trials has reached more than 95%.Xu Ruihua believes that China's existing cancer screening technology still has many limitations, and the rate of early diagnosis of cancer is only about 20%.Artificial intelligence has broad application prospects in early diagnosis and treatment of gastric cancer, lung cancer, breast cancer and liver cancer.

　　- Helping public health science decision-making. Industry experts believe that artificial intelligence through the massive data simulation of medical procedures, medical diagnosis, medical advice and treatment programs will promote the development of public health policy more scientific.

　　Will "artificial intelligence doctors" replace human doctors?

　　How does “artificial intelligence doctor” “think”? In the case of chronic disease management, the seemingly simple “percentage” actually has a whole set of algorithm models.

　　The fourth paradigm founder, Dai Wenyuan, said that for deep learning, the amount of data for chronic diseases is relatively small, and there may be only 10,000 data samples.Therefore, in the risk prediction of diabetes, the algorithm applies migration learning, semi-supervised learning and interpretable machine learning.

　　“Artificial intelligence will not replace doctors, but doctors who know artificial intelligence may replace doctors who don’t understand artificial intelligence.”Shen Dinggang, co-CEO of Lian Ying Intelligence, believes that the application of artificial intelligence in the future will run through the entire clinical workflow, from imaging at source to later diagnosis, treatment and evaluation.

　　However, with current technical limitations, there is often a hidden layer between the data entered by artificial intelligence and the answer to its output, known as the "black box."The consequence of the "black box" is that it is difficult to judge whether the artificial intelligence is wrong.

　　"If you can let doctors see what the computer thinks and how to draw conclusions, it will enable humans to trust the computer more and make humans more reassured about it." Zhang Kang said.

　　Xia Huimin said that artificial intelligence does not replace doctors, but it can greatly reduce the workload of doctors.

　　。She said that better technical means and platforms can not only solve the problem of insufficient medical service capacity to a certain extent, but also improve the fairness of health services.

　　When will the “Artificial Intelligence Doctor” be “on duty”?

　　The reporter visited a number of hospitals and artificial intelligence companies and found that there are many difficulties in the development of medical artificial intelligence at this stage, such as the difficulty in obtaining medical data, the long development cycle, high clinical trial costs, and high hospital application thresholds.

　　Experts believe that China has great advantages in the number of hospital cases, but because medical data is not shared, there is an "island" phenomenon.And the lack of data entry standards, resulting in a large number of high-quality data can not serve the development of medical artificial intelligence.

　　When will the “Artificial Intelligence Doctor” be officially “on duty”?

　　Last year, the former State Food and Drug Administration issued a new edition of the Catalogue of Medical Devices, adding a category corresponding to artificial intelligence-assisted diagnosis, which is embodied in the analysis and processing of medical images and pathological images.This means that if medical imaging artificial intelligence products want to take the hospital procurement route, they must pass the corresponding certification.

　　Yang Zhiming believes that artificial intelligence technology has not yet reached universal artificial intelligence, and is still in the "weak artificial intelligence" stage of vertical field. It is still far from mature in terms of reasoning, comprehensive decision-making, cross-domain, cross-species, self-innovation, etc. It takes a long time to replace a human doctor.