On August 31st, the AI+ Medical Sub-forum, one of the important forums of “2018 (Shanghai) China Artificial Intelligence Exhibition and OFweek (Second) Artificial Intelligence Industry Conferenceâ€, was successfully held in Shanghai. This special forum discusses the difficulties and challenges encountered in the commercialization of AI medical technology around AI+ medical technology applications.
In order to solve the current problems of low screening rate, sensitivity, poor specificity, unbalanced medical resources, high cost of medical AI products, and difficulty in landing, the thinking of strategic operations director Zhang Yue showed more based on the current cervical cancer screening. The AI ​​cervical pathology cytology-assisted screening robot iDeepWise developed by the modal human-computer interaction engine technology.
Deep Thinking Strategic Operations Director Zhang Yue
According to Zhang Yue, iDeepWise realized TCT cervical cytology screening and cervical cytology-assisted diagnosis through deep learning, migration learning and other artificial intelligence technologies, thereby improving the efficiency of cervical cancer screening, reducing the workload and work of pathologists. Strength, effectively reduce the chance of misdiagnosis and missed diagnosis.
Zhang Yue also talked about the company's vision for the future to provide artificial intelligence health consultation and major disease screening services for 300 million families, and to create large-scale crowd health consultation and major disease screening entrance, contributing to individuals, society, and even the country. Belongs to the power of the company.
Although AI medical treatment can not achieve real landing, and encountered many problems in the development of commercialization, in the process, industry people discuss and discuss together, can accelerate the commercialization of AI medical treatment.
The following is the live speech of Director Zhang Yue. The OFweek Medical Technology Network has organized and edited without changing the original intention:
Hello everyone! I am from Zhang Yue, who is deeply thinking about artificial intelligence. The theme of my speech is the healthy dual drive to create a multimodal AI medical brain. Deep thinking team focused on multimodality and understanding and terminal artificial technology, dedicated to solving the pain points and just-needed problems in the medical industry . The team currently consists of scientists in the direction of artificial intelligence and business experts in the medical field, and currently has 34 invention patents. Regarding the problems in the medical industry, our team did a detailed study. As far as the "two cancer screening" is concerned, the female population in China has exceeded 600 million. Among them, the "two cancer screening" females also have more than 400 million, and the probability of "two cancer screening" is only 30%. 80% coverage from the country is still far apart. Let's make a few enumerations on a few of these issues.
First, there are fewer readers. The ratio of Chinese pathologists to population is about 1 to 70,000. There is also a long training period, and the training period for independent doctors takes about eight years. There is also a low efficiency of reading, because the sensitivity of the human eye is only 65%. Then the foreign products of the same type adopt the algorithm of the previous generation, the accuracy is low, and there is the problem of imbalance of medical resources in China. Therefore, the high cost and difficulty of landing medical AI products has become a common problem. In order to solve the above problems, we deeply think about the capabilities of artificial intelligence.
First of all, at the technical level, our basic technology - multi-modal depth and understanding technology is the only AI technology in the field of Chinese human-computer interaction with free cross-domain, context understanding and interaction. We pioneered the ICDCN algorithm based on the deep convolutional neuron network algorithm, which solved the problem of intent recognition with an accuracy rate of 93%. In the 2017 and 2018 SMP Human-Computer Interaction Competition, we won two national championships. In the 2018 Machine Comprehension Reading Competition, we ranked among the top three in the world among the 1,000 teams that participated in the world.
A large part of this is based on image semantic understanding techniques. What image semantic understanding technology? For now, artificial intelligence technology has taken a big step forward. We can not only understand the semantics of speech and text, but also understand the semantics of images. What is the speech understanding of images? For example, there is a picture of a dog and a grape rack. Using traditional recognition methods, we can identify a dog and a grape rack in the picture, and use the semantic understanding of the image to identify this. Pictures, you can read a hot day, a dog under the grape rack to enjoy the cool, this is the semantic understanding of the image, we also use this technology to achieve the recognition of glandular abnormalities.
Most of the abnormal glandular cells are gangs. We not only have to divide cells and identify cells. It is also necessary to understand the arrangement and arrangement of gland cells. Our products are also the only products in China that can identify abnormal glandular cells. Our classification of cells has reached 99.3% accuracy. We are 1% higher than the National Institutes of Health in the United States, with a specificity of 1% higher and a sensitivity of 1.5%. In the 2017 kaggle precision medicine competition, we also ranked first in the Chinese mainland team, when the total number of teams participating in the world was more than 1,300.
The world's first AI chip in the medical field, which we launched in March this year, currently has five national invention patents. Now we have 1.2 million copies of cervical cell sample data, which is our data advantage. Since our products currently cover 60%-70% of third-party testing organizations, the data in the data center has grown by an average of nearly a thousand. Regarding the quality of the data, we have set up a three-layer quality control system, and there are top experts to control the quality of the data. In addition to this, we have added a semi-supervised automatic labeling method for machine learning, which is pre-labeled by the machine, and then the experts perform calibration. Our products are not only able to adapt to the pattern and the method of sedimentation, like the centrifugal sepals can also be adapted. Based on the advantages of the above technologies and the advantages of the data, we propose a solution in which the health screening module and the health consultation module are dual-driven to create the AI ​​medical brain.
The health screening module is currently cut from cervical cancer, and then gradually covers breast cancer, immunohistochemistry, and auxiliary screening of pathological cells such as prostate, providing an auxiliary screening service for pathologists. Our health consultation module mainly uses artificial intelligence technology to learn the knowledge of famous doctors, copy famous doctors, and then provide health consultation for users, as well as artificial intelligence consulting services after pre-diagnosis. A new model for cervical cancer screening is achieved through a health module solution, and a primary screening plus shunt is performed in one test. At present, this sensitivity reaches 98.4% and the specificity also reaches 99.7%.
Now our first step is to focus on the rate of vaginal discharge. At present, a pathologist can read up to a hundred films a day. By combining human and computer, we will screen out most of the negative films first, and then let the doctors spend more time and energy. Focus on positive and suspected positive pictures.
Our products not only improve the efficiency of screening, accuracy, but also universality. At present, our application adopts two modes. The first one is the cloud mode, which is to access the AI ​​brain cloud through the smart device of the access terminal, thus providing AI auxiliary screening and health consultation services. For example, the pathological scanner is connected to the AI ​​brain cloud, and while providing a supplementary reading service for the pathologist, a large amount of data is transmitted to the AI ​​brain cloud.
At present, our health consultation service has been connected to Xiaomi speakers, and will soon become the provider of mobile phone AI health consulting services for Huawei's new mobile phones. We have learned more than 30 well-known experts through artificial intelligence technology, and then we will continue to learn 100 authoritative experts. WeChat public of the national maternal and child field expert Duan Tao has now accessed our artificial intelligence health consultation service, and Jiuyang's healthy kitchen products are also connected to our recommended healthy recipes and nutritional meals. Our vision is to provide artificial intelligence health consultation and major disease screening services for 300 million families, and to create large-scale population health consultation and major disease screening portals. My speech is here, thank you all.
Rabies Vaccine After Animal Bite
Rabies is a viral infection transmitted by saliva of infected mammals, most commonly wild, terrestrial carnivores (e.g., skunks, raccoons, foxes, coyotes) or bats.rabies virus usually moves along a neural pathway and enters the CNS. After entrance into the CNS, the virus is unlikely to be affected by antirabies antibodies and encephalomyelitis usually develops and almost always is fatal. Vaccination of rabies vaccine in a timely manner is mandatory.
Rabies Vaccine After Animal Bite,Animal Bite Rabies Vaccine,Freeze Animal Bite Rabies Vaccine,Freeze Rabies Vaccine After Animal Bite
Changchun Zhuoyi Biological Co., Ltd , https://www.zhuoyibiological.com