At present, artificial intelligence medical imaging products are mainly used in the stage of disease screening, which is an area which is not easy to obtain stable business model. There are at least three preconditions for the commercial success of an artificial intelligent medical imaging product in the future.
Does artificial intelligence make money in the field of medical imaging?
The answer is yes. In terms of market size, the industry report predicts that China's medical image market will be about 600 billion to 800 billion yuan by 2020. But the problem is that there is no such thing as an artificial intelligence project.
No mature business model is the biggest problem plaguing the industry. Although in recent years, artificial intelligence, medical imaging project one by one, even the giants like tencent, baidu, ali also successively into the battlefield, but did not put forward a project themselves in artificial intelligence are profitable medical imaging project. Does the path that relies on screening, analysis and reporting can't get through? It also takes time to verify.
Who is the buyer of artificial intelligence medical image? Who is the real user of artificial intelligence + medical imaging? What is the business model for this field? With these questions, let's explore the difficulties of the commercialization of artificial intelligence medical imaging.
AI+ image: only burning money and not making money?
Founder securities industry finance director, head of the medical industry investment m&a tianjiao ginger in 2017, the Yangtze river industrial BBS (fall) and medical health data and artificial intelligence conference pointed out that at present, most of the artificial intelligence has achieved products fall to the ground, but the real realization of business model to the ground is auxiliary diagnosis and treatment, pharmaceutical research and development and gene sequencing products, image read piece still stop in product landing stage.
Admittedly, the use of artificial intelligence in medicine is a profitable case. IBM Watson is a project that has entered the testing phase.
At present, Watson for Oncology assisted diagnosis and treatment solution has been in China, the United States, the Netherlands, South Korea, Thailand and India and other countries. For its operation mode in China, IBM choice through the distributor to open the market, in this way, you can gain a certain income, according to the current its every time not less than 5000 yuan per person in China charge standards, profit targets.
Let's take a look at what makes Dr Watson profitable. First, IBM as a world famous company, its products around the world, including Asia and Europe and other countries market is recognized for its bigger market provides its business model on the basis of the. Secondly, IBM itself information analysis, natural language processing and self learning ability is very strong, in the aspect of data cooperation memorial Sloan caitlin cancer center in New York (MSCKCC) is well-known medical institutions tumor field, a leading technology and data. Furthermore, Watson's primary application and diagnosis of cancer is not a screening area. In other words, Watson's solution must be able to solve practical problems and have strong success stories and corporate backgrounds as endorsements.
In short, Watson to doctors recommend the best treatment plan, improve the accuracy of the doctors, compared with CT is comparatively mature methods such as field of screening, diagnosis and treatment process of auxiliary for doctors and medical institutions more attractive. Medical institutions are more willing to pay for products like IBM.
However, most of the domestic ai medical imaging projects do not have such conditions.
First, most of the domestic ai imaging projects are focused on the stage of disease screening, mainly by reading the film to determine whether the user is sick or not. While it can reduce the amount of doctors' work, it can improve the accuracy of doctors' diagnosis, but it is not a new demand for medical institutions.
In addition, screening for disease is a costly thing. Zhang jinglei, vice President of Airdoc, an artificial intelligence enterprise in the medical field, points out that the target of disease screening is a large group. For example, a total of 353,000 people in haidian district of Beijing are screened for an eye disease, and at least 1,000 screening sites are needed, and it takes a long time, and the cost is very large.
Secondly, although the Chinese medical image project of artificial intelligence of accuracy claims to remain at around 97%, but health point reporter found after visiting too many related companies, these data is part of the project from the hospital, part of public information or database, in the future, these projects need to absorb more data to improve the algorithm. Professor of fudan university institute of medical image intelligent diagnosis Liu Lei had mentioned in an interview with the media, the current Chinese clinical data comparison, error, loss, cause artificial intelligence of medical data processing is not very successful. "Eye for example, diabetic retinopathy screening, 100, 000 videos on the Internet can be downloaded for free, but there are good and bad movies."
In addition, if companies want to get data from hospitals, they need to reach a mutually beneficial cooperation model with hospitals, which can't achieve profit without collecting money or receiving less money. At present, in the medical image segmentation market, China's more than a large number of high quality data on dimethyl hospital, the characteristics of these medical institutions is ability, capital scale, but do you have any purchase desire of artificial intelligence, medical imaging products is also up for debate.
Tsinghua university, tsinghua university - Qingdao data science institute executive vice dean Han Yishun also in 2017, the Yangtze river industrial BBS (fall) and medical health data and artificial intelligence conference pointed out that artificial intelligence project and there is contradiction between medical institutions and medical institutions to acceptance of artificial intelligence is not high.
"Artificial intelligence people say I help hospitals with screening, and medical institutions will consider screening for the wrong people. Artificial intelligence says I can improve the efficiency of the doctor, the doctor will say that the efficiency has what good to me?" Han yishun emphasized that artificial intelligence requires more understanding of doctors and doctors to find a way out.
It is not hard to see that the artificial intelligence medical imaging project is still in the burning stage, but can each project burn the money? Let's take a look at what are the companies that currently do artificial intelligence medical imaging projects in China?
The first type, BAT, which has a lot of money. They are characterized by money and technology. Second, equipment companies such as wingspan technology, xi 'an yinggu technology, etc. Such companies are characterized by the ability to rely on medical imaging devices or image management collaboration systems to access medical institutions and to obtain data as well as financial revenues. Third, technology companies, such as the aforementioned Airdoc, such as hui hui, etc. These companies have their own algorithms and have an early layout in the field of artificial intelligence, which is more accurate in their accuracy.
Among these three types of enterprises, the first class can rely on their own strong financial support for project development. The second type can rely on revenue from other businesses to support the ai business. A third group relies on financing and a small income to support business development.
The grass-roots level is artificial intelligence main battlefield?
At present, the main artificial intelligence medical imaging products on the market are mainly focused on the screening of several diseases, such as esophageal cancer, lung cancer, sugarnet disease, cervical cancer and breast cancer. But not every disease is screened by artificial intelligence. In the case of pulmonary nodules, zhang jinglei told the health reporter that Airdoc has developed a pulmonary nodular recognition algorithm based on massive data and it is difficult to land in the process of promotion.
On the one hand, the identification of cancer is about the health of a person's life, the medical system and the patient will be very cautious. In real life, on the other hand, a single algorithm to be born, one of the images may have a variety of pathological changes, but the recognition algorithm of lung nodules only able to identify a single disease, it can not solve the actual demand of the doctor.
So which departments are more likely to apply artificial intelligence? Zhang jinglei pointed out that the premise of the commercial success of an artificial intelligence medical imaging product is three:
1. Intensive demand in the department.
In other words, there is a shortage of doctors. Since the past few decades, the medical imaging has developed rapidly in our country, but insufficient resources of specialized subject doctor, and mainly concentrated in big cities and large hospital, many small and medium-sized cities don't have enough ability of imaging diagnosis, many patients to get better medical resources, have to go to big cities.
In ophthalmology, for example, in 2016 the Chinese medical association 21 times, according to the data published academic conference of the national eye at present China has 20% of the county-level hospitals had no eye, and compared Chinese ophthalmologists population is less, especially in ophthalmology medical professional more scarce. According to the statistics of the Chinese medical association ophthalmological branch, there are about 28,000 ophthalmologists in China, but there are fewer than 4,000 patients who can perform endoocular surgery, and the resources of ophthalmologists are still lacking. Meanwhile, the distribution of ophthalmologists is extremely uneven. 70% of ophthalmologists in China are distributed in large and medium-sized cities, and the number of primary ophthalmologists is small.
Airdoc, therefore, the current main products is diabetic retinopathy auxiliary analysis system, the system is used to identify screening sugar lesions, at present, Airdoc artificial intelligence auxiliary analysis programs have been many, including Shanghai changzheng hospital, the hospital has got popularity.
Ali health, tencent and Intel also pay great attention to the selection of the department. "Doctor You", an AI medical product released in July 2017, will be applied to the CT lung nodular intelligence test. Tencent released in August 2017 the first "tencent find shadow" AI medical imaging products, auxiliary of early esophageal cancer and other disease screening, in addition the system also involves the lung, sugar disease, cervical cancer and breast cancer and other diseases. In early 2017, Intel released an auxiliary diagnostic system for thyroid nodules, based on ultrasound imaging. The number of doctors with these diseases is relatively small, and the task is heavy. Artificial intelligence has a large application space.
2. The incidence of disease in the population is not low.
Mr Zhang stressed that, as a product used in the screening phase, artificial intelligence needs to play a role in screening. If a disease is so low in the population that it can only screen one patient out of 10 million people, the cost of screening is too high for commercialization.
3. The subsequent complications were serious.
Or in the eye, for example, diabetic retinopathy is a kind of serious diabetes complications, 30% of diabetics have diabetic retinopathy, and patients with the disease is likely to lead to blindness, if early detection, early treatment effectively, can avoid the occurrence of blindness.
From the above summary of the characteristics of the department, at present, the basic medical market has more demand for artificial intelligence medical image. On the one hand, there are experts in sanjia hospital and advanced technology and equipment, and their demand for artificial intelligence is not very large. Grassroots medical staff, on the other hand, ability is insufficient, but carries a large number of common disease, frequently-occurring disease, chronic disease diagnosis and treatment work, they instead more need to improve the level of diagnosis and treatment and the efficiency of the tools.
Artificial intelligence, which is high in accuracy and high in detection efficiency, is obviously more suitable for primary medical care than artificial ones. However, there are still some difficulties in the field of artificial intelligence medical imaging. There are three main reasons, one is to promote the products to the grass-roots level less. Second, the basic hospital IT system conditions are poor, IT is not convenient to connect with AI products. The third is the high price of artificial intelligence medical imaging products, which can hardly cost tens of thousands of yuan or tens of thousands of yuan to buy a software.
In the grassroots medical institutions have a layout, and cash flow test in accordance with the figure of medical told health points, in a case of successful operations, basic-level hospitals are likely to buy artificial intelligence products. According to the figure of medical vice President Cathy Fang explained, at present, in accordance with the figure of medical in the process of popularized to grassroots mainly there are two ways, one is to provide products to 3 armour hospital, 3 armour hospital radiation by the grassroots medical institutions in 3 armour hospital in the process of learning, can feel the advantage of the products. Second, through agency channels for sales. However, it is worth noting that the charging model of its products is still in the exploration stage, and it is possible to purchase, lease or telemedicine.
How far is AI from earning money?
Burn money easy to make money difficult, artificial intelligence medical image really can't walk a business model? Zhang Jinglei gives health point press a positive answer: "the industry is now actively explore business model, with data accumulated more and more, algorithm is more and more mature, business model will be more and more clear."
Zhang jinglei made such a possibility. In the future, as the technology matures, the cost of artificial intelligence medical imaging products will be lower and lower. This makes it easier for products to open up and be accepted by medical institutions. In addition, artificial intelligence medical imaging products are mainly used in the stage of disease screening. Screening for disease can reduce the chance of a major illness, and thus lower health care costs for the government. The government is also likely to pay for it.
In addition, it is also a way to generate more value in the back. In the future, ai will do more than simply repeat screenings, but do more instructional work, such as developing the IBM Watson diagnosis and treatment program into different diseases.
Overall, although the current business model of artificial intelligence medical imaging products is not yet mature, the prospect of this field is very significant. Once the business model has gone through, artificial intelligence medical images could become a hot vents like a Shared bicycle.
|
