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.
With the rapid development of big data, cloud computing, and the Internet of Things, the application of artificial intelligence technology in the medical field has become more common, playing a more important role in the medical industry . At present, the policy is extremely beneficial to the continuation of artificial intelligence. Governments at all levels continue to subsidize AI, allowing more local products to achieve medical AI cornering overtaking.
In this AI+ medical sub-forum, Dou Yumei, senior director of Yitu Medical, launched the clinical research intelligent solution of the disease library from the cases of intelligent auxiliary diagnosis systems, and carried out financing with the support of the current government, through intelligent medical image-assisted diagnosis, Internet + smart medical and medical big data intelligence have realized the landing of AI medical application field and promoted the development of product commodification.
In the application of medical AI, the practice of landing and product commercialization are the focus of future work of Yitu Medical Technology . Dou Yumei said that in the future, more intelligent solutions will be launched for various cases to help the medical industry. At the same time, we hope to gather more and more people of insight, excellent talents, and talents with an international vision to build a new era of AI and build a new AI medical industry.
The following is the live speech of Dou Yumei, senior director of Yitu Medical . The OFweek Medical Technology Network has compiled and edited without changing the original intention:
I am very happy to have this opportunity to talk to you today. Today I will talk about AI medical from the perspective of the company, and the theme will be more industrial. At present, AI Medical has been developing for more than two years. From the laboratory research stage to the market, it has become a commoditization and industry. What practices and new knowledge have the enterprises obtained?
The doctor is a philosophical job. This is a point of view that we are awesome from the bottom of our mind after two years of entering the medical field. The concept of AI medical care, I feel very much like the concept of evidence-based medicine, combining three elements:
First, the doctor's own experience.
Second, the patient's personal wishes.
Third, scientific evidence and data based on the reality of the objective world.
I think AI Healthcare is also a concentration of these three concepts.
Throughout the development of these two years, in fact, AI has brought about an ups and downs impact on the new development of socialization of society, science and technology and all aspects. This concept is impact-oriented. Workers in the field of science and technology will find this concept, the frontier of this technology, and bring great challenges to the research work and product development work of each of us.
Our company has actually been established for more than four years, initially we did computer vision work. At the end of 16 we entered the medical field and started the technical troika: one is the vision of the computer; the second is the natural semantic understanding, which is the structure of the text; the third is the speech recognition. These three main core artificial intelligence technologies actually light up the number of products we have now.
Throughout the course of more than two years, we have experienced an unparalleled era. In February this year, Professor Zhang Kang from North America published a paper on the diagnosis of fundus diseases using the original pneumonia algorithm model. The impact on the entire AI medical treatment is very large, which makes us aware of this industry again. Refresh, let every AI medical company be more cautious and serious in the process of doing products and doing research.
The impact of AI on the medical field may be summarized in only eight sections: from imaging, in-depth to clinical applications, to disease banks, MDT multidisciplinary diagnosis and diagnosis, and basic research, pharmaceuticals, and drug-based drugs. Mining and other basic research, in fact, AI can really bring a lot of impact and technical subversion to multiple sectors of medicine.
Let's talk about some of our own explorations in this field. We are now involved in three aspects, the first is our starting point - the diagnosis of images. Why should AI medical enterprise development begin with this diagnosis of pulmonary nodules? The reason is that the lung nodules have a large data set on the open network, which is very convenient for the startup company to design, verify and train the algorithm model. But this is only a very, very small starting point.
The second is the intelligent Internet medical platform. In this field, we are exploring the pediatrics, the comprehensive diagnosis system for the pediatric children's common diseases before and after the visit. In the mining of medical big data, we have the intelligence of medical big data. Wait a minute, I will introduce you to our staged achievements in this field, and it is the common practice with Huaxi Hospital.
First introduce the representative products of the image. We have been through more than two years from the lung nodules two years ago to the third-generation image-assisted diagnosis system for lung cancer today. This product has now entered nearly a hundred hospitals, some embedded in the workflow, and some in the imaging department to do product trials. At present, the third generation of lung cancer products have been applied to the benign and malignant judgment of tumors. The advantage in lung cancer imaging diagnostic technology is our core technology advantage. We have a team of more than 300 doctors and medical students to give us gold labels.
After more than two years, we have become closer to the real needs of clinicians in the quenching of algorithmic models. At the same time, we use data above the million-level data, which may be the largest data set for the use of internationally, training lung nodules, lung cancer, and tumor imaging diagnosis.
The second imaging product is the intelligent imaging diagnosis of the mammography target we released this spring. This is a product we have worked with Fudan Cancer Hospital. Mammography's molybdenum target products have several leading points in the global scope. One is the design of the algorithm and the detection of the lesions, including some new algorithms proposed in 3D reconstruction.
In terms of children, we have an intelligent diagnosis of the bone age of children, a specific application in the image. By scanning the child's left-hand X-rays, and then in the second time, you can intelligently determine the age of the child's growth and development, and even accurate to the natural moon. The purpose of this product is to be the first bone age library in China that belongs to Chinese children. Because Chinese children do not have a bone age map, we must use this to be the Chinese nation's own children's bone age library.
On the application side of the intelligent Internet medical platform, our first exploration is the pediatric aspect. Because the resource allocation of Chinese pediatricians is very uneven, there are probably hundreds of thousands of pediatric doctors, so we chose the field of pediatrics. The current system covers two parts, one part is the B-end - the hospital side, on each doctor's own workbench. The second end is the C-end - the client, on each user's own mobile phone. At present, we use the WeChat applet on the mobile terminal of the mobile phone, so pre-installation is very easy, and almost no pre-installation is required.
This system runs through three different stages before, during and after the visit. First, the role of AI is the nurse station, then the doctor assistant, the convenience service, and his own expert health steward. It has truly created a patient- and child-centered experience process for medical treatment.
One of our solutions for medical big data governance intelligence is the research big data platform of the disease database. On this type of platform, we have all the data on the hospital's overall information system, and then put it on this platform to do big data governance. And we propose a series of artificial intelligence model algorithms, aiming at clinical problems, algorithms and strategies for scientific research, and ultimately create a series of scientific research applications that doctors need.
In the field of medical big data intelligence, our first phased result is the MDT diagnostic system for lung cancer single disease control jointly issued by Huaxi Hospital. Its core competence and core technology lies in the full extraction of complex medical semantics. This algorithm model is very complex and is a combination of a series of models.
Together with Huaxi Hospital, we released MDT's intelligent diagnostic system, which covers almost all the data of lung cancer patients in West China Hospital over the past decade. Our most important job now is to develop a comprehensive MDT decision-making advice module.
I have just brought together some existing and mature products from Yitu Medical that can go from the laboratory stage to the real clinical and go to the hospital and department. What kind of difficulties do we really have at the stage of industrialization? Or where are our boundary conditions? We have listed a few points.
The first boundary condition, or the ceiling of the entire medical industry, should be the policy compliance of the state and industry.
The second boundary condition is the establishment of a business model, which also determines the success or failure of a business.
Secondly, the financing conditions of the overall environment, in the overall medical industry have a double ten statement, especially for pharmaceutical companies, that is, the first decade of investment, 10 billion out of formal products. In the AI ​​medical field, we believe that the three to five year period is a phase of heavy investment, so financing is also very important for us.
Then let me dismantle these three elements. The first one is policy and compliance. At present, in the international arena, China and the United States are truly achieving a win-win situation in AI innovation. Therefore, from the perspective of the international scope of the United States, the United States has gradually approved 10 approvals for the approval of medical applications for artificial intelligence. The latest approval is an application for intelligent diagnosis of cerebral hemorrhage and then to help doctors make recommendations for triage. In fact, such a recent application also reveals from a single perspective the two meanings that AI applications bring to doctors.
First, it involves the decision-making of the doctor's auxiliary diagnosis.
The second is to closely fit the doctor's own workflow.
If the AI ​​application is all in the doctor's own workbench workflow, then the responsibility identification and risk are actually a problem that cannot be circumvented and must be faced positively.
In this regard, there are recent regulations in Japan, such as Japan, which clearly stipulate that AI applications are positioned as doctors' auxiliary systems, and the ultimate responsibility is determined by doctors. So AI applications are endless devices, applications, and their role is ultimately an auxiliary function, and it is not possible to really replace the doctor. In China, AI as a national development strategy, the entire country's policy drive is huge, so on the whole, China's development speed will be faster than other countries, the United States and Japan. Including the overall layout of China's overall AI application in 2030 and requirements.
From an approval perspective, CFDA is only more rigorous than the FDA. Currently, in the new medical device catalogue, the vast majority of AI applications are targeted to the third category of medical device applications. Therefore, the requirements for clinical trials are very strict, which requires us to be an AI medical enterprise. We must have high requirements for the construction of the accusation system, so that we can finally obtain the CFDA certificate and distribute it to the market. , conduct legal and legal business sales.
At the same time, in terms of business model, it is crucial for companies to build a business model. We still position our customers as hospitals or medical institutions today. In setting up a business approach, we may learn from some traditional medical device software or some business models of pharmaceutical companies, as well as innovative parts.
There is a foresight, there is Watson, I define it as "Explorer of the 0th batch" because it does not belong to the first batch. The first batch is in our historical process and is evolving this historical process. Watson's way of charging is to charge the hospital and the patient at the same time. The contradiction between over-marketing and backward technology has caused Watson to cut 70% of its employees.
At present, our business model is built. If we look at medical institutions as our overall users, we are actually layered. The first layer is the cooperation and mutual benefit of high-end hospitals, that is, the product research is built on the high-end head-top hospital. The basic hospitals are popularized and installed, which means that the primary hospitals will receive the high-quality medical resources and knowledge transfer of these top three hospitals, and solve the serious imbalance of medical resources in China through AI products.
At the same time, for enterprises, a big cost that we can't ignore is the interconnection with the original medical information system. This cost is divided into two parts, one is the time cost, and the other is the real capital cost. If we want the primary hospital to install HP and use AI medical products , we must complete these interconnections in the fastest time.
Finally, let's talk about the financing environment. In the overall global financing environment in 17 years, China surpassed the United States in terms of AI and got 48% of the total global wage. This is quite unexpected. Explain that from the financing conditions and financing environment, China is indeed directly competing with the United States. A deeper analysis of the current financing environment means that the amount of financing is decreasing, the overall amount of financing is increasing, and the pace of financing is slowing, but resources are increasingly flocking to the head players on the head track.
If we start from the identity of the real AI medical institution practitioner explorer, let us look at the position of AI medical care in the context of China today. Our judgment is that China's AI medical industry has entered the stage of industrialization. The industrialization phase will reflect two very core features in one to two years.
The first is to falsify the reality, that is to say, from the speculation and over-marketing to the AI ​​enterprise, the medical enterprise must come up with a practical and able to enter the clinical process to help doctors improve efficiency, reduce missed diagnosis and misdiagnose medical risks. Bring them real, valuable products. The second is resources, the overall medical resources, data resources and expert resources, including the financing of resources, will be concentrated to some of the head players, so this is the two typical characteristics.
This is a brand new era. Because we see it, we believe that we hope to gather more and more people of insight, excellent talents, and talents with an international vision to build a new era of AI to build a new AI medical. Industry, thank you all.
Incubator & Oven Products,Oven Products,Constant-Temperature Incubator,Thermostat For Incubators
Guangdong Widinlsa International Co.Ltd , https://www.guangdongwidinlsa.com