Medicaltranscriptionsservice Imagine: In the days to come, if AI treatment becomes a reality, you can use a computer directly at the hospital to see a doctor. Through laboratory data, the system quickly gave you a diagnostic result that allowed you to prescribe medication and even decide whether or not to undergo surgery. Although this process sounds very convenient and beautiful, you have to reach the utopia of the AI ​​clinic. Actually, it is not that simple, and it is not as easy as it is to imagine whether or not its bricks are under construction. Stay in the lab for AI treatment The user groups for AI diagnosis and treatment are not limited to assisting doctors. According to different service groups, they can be divided into four levels: Respond to patient consultation Assisted guidance staff, pharmacy personnel, and medical online customer service; Clinical assistant decision-making of primary doctors and general practitioners; Patient recovery tracking for doctors. The above functions must be related to AI diagnosis and treatment. If a medical startup company does not have a lot of data, it is impossible to make a mature product. In the absence of data, start-up companies nowadays mainly cooperate with hospitals or data centers to obtain relevant data. The main purpose of these data is not yet used for actual diagnosis in hospitals, but only for the accumulation of the previous period - optimization of algorithms through massive clinical data. Jeremy Howard, founder of artificial intelligence diagnostic Nova Enlitic, once stated that lack of data is the main reason that restricts their development. And last year it received a $10 million funding from Capitol Health to improve diagnostic algorithms through radiology data provided by the latter. Through the excavation of massive medical data, real-time accurate diagnosis of medical images is realized, and the probability of illness is predicted, and doctors' decisions are assisted. The most successful non-IBM Watson is currently the most successful AI diagnosis and treatment system. According to Lin Xueting, a software engineer at the Watson Health Cloud of the Tokyo System and Software Development Institute, to the introduction of Lei Feng.com, the system needs a research center that can cooperate to obtain accurate data sources. Watson's practice of acquiring data is mainly to cooperate with CVS, the second-largest chain pharmacy in the United States, to obtain user behavioral information, clinical data, drug purchase data, and insurance information, or to cooperate with laboratories and research centers around the world. However, the current research work is mainly limited to the laboratory. Even with Watson's successful case of diagnosing leukemia, it is still a long way from clinical trials. Hsnewsbeat Validity of data The biggest flaw in AI diagnosis and treatment is the data collected for medical research institutes, and there is still a distance from the real question. One reason is the small size of the data. The current practice of most technology companies is to exchange customized solutions by obtaining data from hospital agencies. However, these clinical data can only be considered "little tricks." Lei Feng network simply for everyone to count: In the United Kingdom alone, there are nearly 200 people every day can not see tomorrow's British rain (handy black). By 2020, 200 million people will be blinded globally due to AMD, a retinopathy secondary to diabetes. However, even with the combination of Google DeepMind and Moorfields, a century-old hospital in the United Kingdom, the current training data can only reach more than 1 million anonymous scans. In this comparison, the data provided by independent hospitals is simply a drop in the ocean compared with patients worldwide. At the same time, the acquisition of data on the disease will be limited by the region and even the disease, which makes the validity of the data a discount. Another problem is that the quality of the data needs to be improved. Insufficient e-degrees, poor data collection methods, lack of standard systems, and data sources with low levels of structure have made clinical trials difficult. Medical data is not as granular as financial data, and its degree of granularity and professionalism are relatively mature. At present, the level of HIS and EMR in hospitals is far from enough. "The quality of data is the basis for effective analysis. At present, data cleaning work takes up too much work, and in the end it is a quality problem," said Zheng Jieru, CEO of Shulan Hospital. He believes that the average age of doctors who use hospital information systems is more repellent to the latest information systems, and there is no urgent need for data analysis. Therefore, it is difficult to establish excellent data structures and data. Quality". Lei Fengwang asked several doctors in the first- and second-tier hospitals. They said that the hospital has not introduced relevant artificial intelligence diagnostic facilities because the accuracy rate is not optimistic and is still in a wait-and-see mode. However, it does not rule out that “when the regulations permit, it will Artificial intelligence diagnosis as an auxiliary diagnosis." Kang Fuzi CEO Zhang Chao said to Lei Fengwang (searching for "Lei Feng Net" public number) that "the current market diagnosis (doing many years of expert system) is mainly based on symptoms, a few can add laboratory data, but in fact History, medication, incentives, etc. all need to be learned step by step.†Assay data are more often used as a reference for the moment, and doctors’ “expectations†are for symptoms, incentives, history, medication history, etc. Comprehensive consideration of dimensions. The poor quality of data will inevitably hinder the learning of artificial intelligence. In addition to quantity and quality, the absence of law has also questioned the validity of the data. Regardless of the level of informatization of these data quality, not to mention the fact that DeepMind caused public opinion attacks because it had reached an annual data exchange of 1.6 million with the NHS. In the latest specification, Apple also does not allow developers to store data on iCloud. It is also a specification that technology companies use to avoid the risks associated with leaking data. As of now, there are no relevant laws and regulations related to AI treatment at home and abroad, and there is no clear norm for the relevant responsible subjects and the procedure for the treatment. At present, third-party foreign countries can only use personal data statistics according to the HIPPA agreement. This medical insurance carrier and liability law, enacted in 1996, aims to protect the privacy and health-related electronic data of patients and to standardize the data exchange process as much as possible. The technical guarantee defined by HIPAA's security principles does not require the use of a specific technology, but an adjustable framework that requires organizations to adopt appropriate technologies as much as possible to protect data security. These security solutions need to be implemented. "Check control, information integrity, data transmission, etc." Various requirements. Because of privacy concerns, medical data is usually impossible to share on a large scale, but it is contrary to people’s natural rejection of the disease. Naturally, it is unwilling to “share†with other people's laboratory data. The problem of information silos has exacerbated this situation. Venturebeat summary Simply put, the main reason why AI's diagnosis and treatment fail to develop rapidly is that the quantity and quality of data are not sufficient at the current stage to support the actual questioning and treatment. The diagnosis and treatment is a very individualized and personalized activity. The large-scale opening and application of medical data is required. Besides the analysis of big data based on semantic natural language processing, it also needs legal support and protection. In this way, medical big data can truly serve the exploration of artificial intelligence in terms of effectiveness and provide assistance and support for AI diagnosis and treatment. Maybe in the near future, we will be able to reach Utopia and enjoy the convenience of computer access. Recommended reading: Watson diagnosed rare leukemia for the first time. What's difficult to diagnose in 10 minutes? After deepening Go, Deepmind wanted to “see through†human eyes Static Frequency Connerter With Threephase Output Variable Frequency Inverter,Variable Frequency Converter,3 Phase Frequency Converter,Frequency Converter 60Hz To 50Hz 3 Phase Jinan Xinyuhua Energy Technology Co.,Ltd , https://www.xyhenergy.com
