Prospects for the development of medical artificial intelligence

Medical Network May 8th, at present, artificial intelligence as a breakthrough point in the development of the global medical field is high hopes. The 13th international medical device regulatory agency forum held in Shanghai recently held a special discussion on 'artificial intelligence medical devices'. .

The industry generally believes that whether medical artificial intelligence can be deeply integrated with medical care and full-scale advancement of production, research, and research is the key to determining whether this wave is to become a bubble or to lead the industry transformation.

Recently, reporters at the Radiology Department of Beijing Friendship Hospital of Capital Medical University saw that the entire waiting area had been filled with seats, and even the walkway was full of patients.

'We should be eight or five nights ago, but even if we don't rest at noon, most doctors will have to go to work every day at 7 or 8 o'clock every night. 'The director of the department, Yang Zhenghan, told reporters that there are currently shortages of imaging doctors in domestic hospitals. The work load of doctors is particularly large. 'Especially after the medical reforms reduced the price of imaging examinations on April 8 last year, taking our department as an example, the daily CT volume has increased from 400 to 600. Because the equipment and personnel have not increased, This means that each imaging doctor's workload has increased by 50%.'

'If we can replace some repetitive diagnoses with artificial intelligence, we can greatly reduce the workload of doctors and improve the quality of diagnostics. ' Yang Zhenghan said that taking pulmonary nodules as an example, after the patient has completed the examination in the hospital, if it can All the lung nodules are discovered and identified by artificial intelligence technology. Even if it is not judged good or bad, about two-thirds of the imaging doctors can be saved.

It is understood that today's artificial intelligence products, especially in medical imaging, have some automatic identification and information extraction functions that can extract things that are difficult for the human eye to distinguish. Although they are not substitutes for doctors, some can provide some doctors. Imperceptible auxiliary information.

Although the level of radiology departments in the top three largest hospitals in China is comparable to that of hospitals in developed countries abroad, the average radiology level is not optimistic on a nationwide scale. In response, Yang Zhenghan stated: 'Whether it is an alternative doctor who simply repeats the work to improve the diagnostic quality, or improve it. The level of diagnosis and treatment in basic hospitals may even help expert doctors to further improve their understanding of the disease. The medical field is worthy of deep involvement in artificial intelligence.

According to reports, the accuracy rate of current medical imaging artificial intelligence products is not the same in different organs and diseases. The current diagnosis of pulmonary nodules, especially pulmonary nodules, is the diagnosis of artificial intelligence. The level is not less than a mature radiologist, and it can almost reach the level of the attending doctor. However, from the point of view of clinical research, there is a problem of poor interpretation, that is, the input data and the output result are correlated but cannot be used. Causal logic to explain.

Zerotek Technology CEO Zhang Tianze admits that the explanatory difference may be accepted in other fields, but in the medical field, it means a great medical risk.

Zhang Tianze stated that there are currently two solutions to the problem of poorly interpreted artificial intelligence in the medical field: First, the introduction of medical experts to conduct further analysis of the causality of relevance, and to give medical artificial intelligence products medical genes, which is equivalent to artificial The causality of the outcome of the smart output presents a hypothesis where clinical experts and technology companies work together to find out the logical relationship. The second is to introduce a large number of real clinical cases and use these cases to explain the results to provide a perceptual explanation for use by doctors. The feeling of successful treatment cases is found in the process of the product. However, this perceptual explanation also requires the specific case that the artificial intelligence technology company has mastered to be reproducible and subject-leading.

'Either Solutions, medical artificial intelligence must be closely integrated with clinical medicine, the need to involve more and more leading clinical experts of high quality, rich dimension sufficient clinical and pathological data.' Zhang Tianze think, alone artificial intelligence company Technology-driven, no medical personnel involved in product development and design, there is no large number of clinical experimental projects and clinical cognition, the product can not be applied in real medical scenarios.

In this regard, Yang Zhenghan high degree of identity: 'Only a doctor involved, especially high-level medical experts, can the product of artificial intelligence Clinical really need to do more in line with practical needs.'

'Artificial intelligence to achieve medical product landing There are no shortcuts and tricks.' Reasoned seats Wei-old president of marketing, science and technology, science and technology in the early AI suppose CT screening for lung assist system development, it is rooted in the hospital, Tongji and Tongji Medical College working closely hospital, Shanghai Changzheng hospital and other medical institutions, and the product iterations, the optimization of the process, together with the medical institutions established product standards, perfect for exploring the robustness, ease of use and safety of the combination.

On April 12, the State Council executive meeting reviewed and approved in principle the "on the promotion of Internet + medical health 'Development Opinions' clearly stated that it is necessary to promote the integration of the Internet and medical and health services in areas such as artificial intelligence applications.

In this regard, the industry generally believe that, from the raw power, China's medical development of artificial intelligence with a bright future, because China has a better policy support, there is a greater source of data, coupled with China's demand for doctors product of artificial intelligence , and so Entrepreneurship The company is also willing to increase investment in rapid development of products.

'For large tertiary-level hospitals and university affiliated hospitals, medical artificial intelligence is an important part of current scientific research work. We attach great importance to enterprise Research cooperation. ' Yang Zhenghan said that the current friendship in Beijing hospital He undertook several clinical research projects for medical artificial intelligence products and jointly undertook several scientific research projects with the Magnetic Resonance School of Capital Medical University. 'If the artificial intelligence company promotes technical capabilities and cooperation in production, research, and research, both parties The cooperation will also increase. '

According to Yang Zhenghan, although most hospitals that cooperate with artificial intelligence companies have not yet entered the substantive payment stage, if there are products that have actually passed the approval of the national regulatory authorities and have excellent clinical applications, I believe many hospitals are very willing to purchase. In the case of the hospital radiology department, imaging equipment is a major cost-bearer, and the second place is the doctor’s labor cost, because imaging doctors are highly educated. Talent Team. Take Beijing Radiology Department of Friendship Hospital as an example. Among the 31 image doctors, there are 21 medical doctors and the rest are all masters.

'If we don't need so many people, or if we have the same staffing, we have more energy to do more technical work, such as investing in research to create more value, and virtually saving resources for the hospital.' Yang Zhenghan Said, 'The application of artificial intelligence products can cooperate with doctors to improve the diagnostic level and reduce the workload. With these two points, I personally think it worthwhile to pay for it.'