2019 has passed, and this year is a crucial year in the history of artificial intelligence development. Since 2014, artificial intelligence has appeared more and more frequently in people’s conversations about technology topics, which has also attracted the enthusiasm of capital. Over the next four years, investor enthusiasm for artificial intelligence grew. By 2018, China’s AI investment and financing scale reached a peak of 118.98 billion yuan. But a year later, the industry has clearly begun to feel cold. In the third quarter of 2019, the investment and financing scale was only 57.717 billion yuan, which was almost cut off.
In the small field of medical artificial intelligence, its development process is basically the same as the development trend of the large field of artificial intelligence. The investment scale has risen from 202 million in 2014 to 8.476 billion in 2018, and the third in 2019. The quarterly decline fell sharply to 3.882 billion, and the amount fell by 50% year-on-year. The cooling of the capital market is because since 2015, many artificial intelligences have copied the Internet thinking to make technologies, but these technologies are not hard-core, resulting in a lot of capital investment but no entry. After experiencing too many bubbles, on the one hand, investors have become cautious. On the other hand, because these technologies are not hard-core enough, they have brought a large number of followers, making it difficult for investors to distinguish the potential of each artificial intelligence project, and they are more conservative. investment direction.
The field of medical artificial intelligence is even more special. At present, even if artificial intelligence is involved in the medical process, artificial intelligence cannot replace the role of doctors. In addition to the deep participation of doctors in AI model training and other links, the learning method of artificial intelligence also has great limitations and cannot be separated from a large number of manual annotations. At the same time, compared with doctors, AI learning still lacks a lot of theoretical knowledge in physiology and pathology, especially lacks the comprehensive utilization of information from different sources, and lacks the same comprehensive judgment ability as doctors. Because of this, knowledge graphs have now become an important breakthrough in the future of medical artificial intelligence research. At the same time, it should be noted that the current medical artificial intelligence is still relatively fragmented and not sufficiently integrated. Different disease models are used in different environments. If a unified integration specification cannot be established, it will become intelligent for the sake of artificial intelligence, and the doctor’s workstation will be transformed into “fragmented”, which will affect the efficiency. Artificial intelligence, the reason doctors often don’t use these new technologies. To solve this problem, only by allowing developers and doctors to mingle and truly understand the overall work of the hospital, can we have a targeted and comprehensive plan.
In 2019, AI applications represented by CDSS (Clinical Assisted Decision Support System) and VTE (Venous Thromboembolism) intelligent prevention and control system were prominent. This is due to the fact that these AI technologies have been embedded in doctor workstations, and even sink into primary medical service institutions. For example, the Lingyi Zhihui CDSS developed by Baidu has been applied to Mafang Community Health Service Center in Pinggu District, Beijing. The doctor’s diagnosis and treatment robot has also been applied to Xiaozhi Town and Shengyuan Community Health Service Center in Pingyin County, Jinan City. At the same time, in the promotion of CDSS, the combination of CDSS and MDT (multidisciplinary joint consultation) is also an important development trend. Because CDSS is based on a single discipline, and the patient’s symptoms may not be solved by a single discipline, it is necessary to embed CDS as a subsystem into MDT to obtain the best diagnostic results and treatment plans. In November 2018, Peking University took the lead and jointly developed the “National Liver Disease and Tumor Multidimensional MDT Artificial Intelligence Collaboration Platform” jointly with the Asia-Pacific Liver Disease Alliance and Yidu Cloud, and implemented it in some hospitals.
In addition to the continuous integration of multi-disciplinary resources by the system platform, the entire medical artificial intelligence industry has also become more and more independent in 2019, forming a trend of highlighting the development of groups and forming some industry alliances. For example, in September 2019, the Edison Digital Medical Intelligence Platform and the Huawei Cloud Medical Health Kunpeng Industry Alliance were established successively. When the environment is cooling down, this kind of group heating is also an effective way to survive the crisis.