The medical world "human-machine war" was unexpected
A few days ago, a special contest was launched in the Sino-Japanese Friendship Hospital. The opponents were 10 dermatologists from Beijing, Yunnan, Inner Mongolia, etc., and the other was the first artificial intelligence-assisted decision-making system for yellow human skin tumors, the Youzhi AI system. The competition subject is the diagnostic success rate of skin tumors, including the nature of skin tumors and the name of the tumor. Before the start of the competition, Professor Meng Rusong of the Chinese People's Liberation Army General Hospital Dermatology Hospital took out 60 sets of skin tumor case data, and randomly selected 10 sets of pictures on the spot to let the doctor and the AI ​​system simultaneously answer. The correct answer is pathological diagnosis as the gold standard. Ten minutes later, the benign classification rate of the Youzhi AI system reached 100%, the malignant compliance rate was 75%, and the average coincidence rate reached 90%. The doctors group corresponded to 76%, 62.5%, and 63%, respectively. AI wins the dermatologist. Meng Rusong said that the results of the "human-machine battle" show that the Youzhi AI system has completed the functional realization of clinical application, and can effectively achieve the clinical work of dermatologists, especially primary dermatologists and low-grade dermatologists. Help. The dermatologist is not surprised that the human doctor is inferior to the AI ​​system. In 2017, Stanford University published a study on the deep learning of skin tumor machines in Nature. The results showed that the coincidence rate of deep learning in the benign and malignant 3 classification and the disease category 9 classification task reached 72.1% and 55.4%, respectively. For the same classification and identification, the average compliance rate of professional doctors was 65.8% and 54.2% respectively. "Because the clinical manifestations of skin diseases are mostly located in the exposed parts of the skin, many diseases can be judged by viewing clinical images, dermoscopic pictures, pathological sections and other imaging data." Professor Zhang Jianzhong, director of the Department of Dermatology, Peking University People's Hospital, said that through the massive pictures Learning, dermatology is the most suitable medical field for telemedicine and artificial intelligence. "For medical institutions at all levels in China, early identification and diagnosis of skin tumors is a challenge, especially for primary hospitals, skin tumors are easily missed, causing great clinical problems." Youzhi AI System Project The lead person, Cui Yong of the Sino-Japanese hospital, said in an interview that the AI, which was launched this time, was developed for this pain point. Although in Cui Yong's view, there are still four major bottlenecks in the current AI system: the dermatological characteristics determine the inaccessibility of absolute big data, the AI ​​skill acquisition path causes the deep learning strategy to be incomplete, and the AI ​​recognition dimension limits the technical path. Maturity and the uncertainty of manual diagnosis affect the accuracy of data resources. However, for grassroots organizations, Youzhi AI will help grassroots doctors improve their diagnostic level and reduce the rate of misdiagnosis and missed diagnosis. At the same time, Cui Yong also hopes to carry out remote consultation services for the difficult diseases that AI can't solve through strategic cooperation with the National Telemedicine and Internet Medical Center. The AI ​​system released this time was jointly developed by the China Population Skin Image Resource Library (CSID) project team and UMG Technology. "The current system is focused on skin tumors, and the next step is to expand to more dermatological diseases, which is suitable for more skin disease-assisted decision-making scenarios." Chang Jiang, CEO of Youmai Technology, said that the technology and the country The strategic cooperation initiated by Telemedicine and the Internet Medical Center is to hope that in the next three to five years, new technologies will be used to help at least 10,000 primary medical institutions and 100,000 primary doctors solve various difficult clinical skin diseases. Lu Qingjun, director of the National Center for Telemedicine and Internet Medicine, also said that the release of the AI ​​system has a milestone in the application of artificial intelligence to the medical field. "Based on the continuous improvement of technologies such as the Internet platform and artificial intelligence, the utilization rate of medical resources will be continuously optimized, so as to achieve a win-win situation for patients, doctors, medical institutions and so on." Distribution Box,Industrial Distribution Box,Electricity Switch Box,Box For Distribution Guangdong Shunde Langzhi Trading CO., Ltd , https://www.langzhielectrical.com