| Micro-channel |
 |
|
|
|
|
|
| |
| Domestic artificial intelligence becomes investment hot spot, which subdivision field deserves attention? |
| |
| Author：中國銘鉉 企劃部 Release Time：2017-8-16 11:21:46 Number Browse：735 |
| |
Medical network on August 16 - the combination of artificial intelligence and medical extraordinary development potential and imagination space, is one of the hotspots of investment by the next five to 10 years, there are a lot of niche is remarkable. With digital imaging diagnostic equipment including conventional X-ray digital equipment, magnetic resonance (MRI), X-ray computer tomography (CT), and medical imaging image management system and its workstation of the popularization of computer hardware platform, the application of artificial intelligence in the field of medical imaging will be the first to enter the fast lane.
The diversity of ai
Artificial intelligence application in the field of medical treatment includes: cases/literature information analysis, medical imaging diagnosis, medicine, speech recognition, virtual assistant, patient risk prediction, hospital management, health management, drug discovery, intelligent medical robots, wearable artificial intelligence and so on more than 10 categories. From the current domestic project, the quantity is not much. According to the incomplete statistics, the total number of projects is over 60, mainly the application of artificial intelligence in medical imaging, and relatively few decision-making products such as drug development.
, according to data from 2006 to 2017, this decade, a total of 26 domestic applying artificial intelligence technology to medical imaging diagnosis, 12 companies will be based in Beijing, four companies headquartered in Shanghai, in both medical talent and resources agglomeration of cities, most of the companies won the favour of capital, expected the next 5 to 10 years have accelerated the trend of growth.
In terms of capital, BAT, with its Internet genes, has long been targeted at the combination of healthcare and AI. For example, alibaba and huarun wandong medical cooperation developed thousands of miles of cloud, invested in landing medicine, ET medical brain; Tencent, as an investor, has invested in carbon cloud intelligence, CloudMedx, and the think network, and launched tencent's search for early screening for cancer. Baidu is mainly on the related patent layout, according to statistics, from 2014 to 2016, baidu to apply for patents, paragraphs 4, 63 and 142, respectively, is not only the number is increasing by leaps and bounds, and covers the artificial intelligence field, deep learning, machine learning, neural networks.
There has not been observed in focus and a large number of layout of medical artificial intelligence investment institutions, dabble in institutions in addition to tencent, ali industries such as funds, mainly some profound Internet gene angel funds, such as very serious fund, investment, intelligent, a doctor and Medal; Sequoia and inspired venture capital invested in hainer medical letters; Sequoia, co-creator and PreAngel invested in the medical association; Feng capital invested in DeepCare; The fund and weft China invest in tuma-deep, and are more cautious about investing in health care.
Secondly, the project is in the early stage and conceptual stage, the business model is immature, and the restriction of medical policies and regulations makes investors cautious. For example, whether the remote imaging consultation platform is currently equipped with the closed loop system for payment, can the diagnosis result be mutually agreed between different medical institutions? For example, how can the input and output of AI solutions be calculated by hospital or third party imaging center? Will hospitals and doctors be motivated? Add artificial intelligence to read piece function, whether the hospital can add price to charge? Who will pay for the incremental change? Is it social security or commercial insurance, or is it self-paying? The expansion and creation of such business models.
In the field of medical artificial intelligence, domestic medical artificial intelligence projects are mainly focused on medical image. Medical image is also a very popular subdivision of foreign countries. But foreign drug discovery and health management, which are also popular abroad, are rarely involved in China. Abroad, for instance, to obtain the total financing amount the top five companies - Flatiron (us $313 million, medical record information analysis), Welltok (us $238 million, medical record information analysis), Benevolent (us $140 million, drug discovery), Butterfly ($100 million, medical imaging diagnosis), Lam Therapeutics (us $98 million, drug discovery) - there are two is to do drug discovery, two is to do medical record information analysis. Many reasons that cause this gap, the main thing perhaps is: artificial intelligence at the input of medical data structured in Europe and the United States needed to medical institutions is very mature, artificial intelligence algorithm update quickly, the output results have medical institutions or commercial medical insurance institutions, can set up a efficient health management closed loop, so diversified artificial intelligence application in the field of health care.
Second focus medical image + AI
Medical image is the focus and focus of domestic AI investment.
There has been a lot of talk about "whether artificial intelligence will replace doctors" in recent years. Indeed, the AI reading is a trend, but does it mean that the imaging doctor is going to lose his job? The author thinks, certainly not, at least in Chinese mainland.
Traditional image diagnosis process is the primary image doctor to write a report, a senior doctors to nuclear, incurable diseases to submit to the director of the weekly briefings, so the diagnosis report issued by time is longer, the fastest also want to two or three days, even the emergency patients with CT diagnosis report, the fastest also want to 1 hour. According to relevant data, the domestic medical image data growth of 30% a year, while the image only 2% increase in the number of physician, through artificial intelligence technology, can reduce the doctor's work load, auxiliary diagnosis to reduce the misdiagnosis and improve the diagnosis and treatment efficiency of the hospital.
But some studies have shown that the accuracy of the AI reading for certain diseases is about 65 percent, which means that the error rate is 35 percent. Therefore, the AI reading can only replace the first step of the imaging diagnosis report by the primary doctor, but the diagnosis and treatment of the nuclear tablets and the difficult diseases will require a high seniority doctor.
If we can fully enter the era of strong ai, the growth path of the imaging doctor may change. The doctors who grew up to be seniority were trained by hundreds of thousands of pieces of reading. After that, the experience of doctors was completely changed. Imaging doctors may be able to take on new roles in the future, such as the growing importance of managing patients' medical data.
From the perspective of investment, why are there more investment institutions focused on medical treatment, and fewer medical imaging AI? The author thinks that there are two main latitude problems to be solved urgently.
First, the payment side. Is it more expensive or cheaper for the hospital image section of the AI? In theory, hospitals should pay less for patients because of increased efficiency. AI's software and hardware, which are paid for by the hospital, have the incentive or incentive to pay for it. Therefore, at least for now, the image of public hospital nuclear radiology, basic did not use additional and incremental read piece of artificial intelligence technology and products, what is used now is equipped with the embedded image recognition and image post-processing software, belong to compare the early computer aided diagnosis of CAD, most don't have a deep learning technology.
Second, the data end. The AI software for medical imaging is universal, a type of machine learning that requires doctors to provide parameters. The key is that the development of AI technology depends on the size of the original sample and directly determines the accuracy of AI reading. Is there enough quality data available to the machine at home? China has a large population and a large data base, but each hospital has its own information, and image data is traditionally a hospital asset, not even a patient. Even if is currently implemented as a 2008 interconnectivity of couplet of Shanghai medical, hospital between read image data, is still in the manner of service scheduling, indicated resource servers to provide data, deep learning should be on the analysis of hospital data, and medical imaging occupies more than 90% of the medical data, but the image is not only a picture, a doctor at the time of interpretation is used more related medical knowledge, the bigger challenge is the structural problems of medical data, although the huge amount of medical data, but IDC Digital, according to a forecast is 80% of the data belong to the unstructured data, prevented the AI precision of the algorithm.
After clarifying the above two questions, we will see which segments to invest in. According to the image generated to the diagnosis of the various stages of the process, medical imaging industry can be divided into: gathering imaging, image archiving, image transmission, image display, image sharing, imaging in the diagnosis of six stages. The author thinks that, the main increment may come from the application of artificial intelligence auxiliary clinical decision, including integrated image diagnostic tools, image recognition, image quality control system, etc.
First of all, the niche market prospects are very optimistic, Markets and Markets predicts that China's medical imaging equipment market scale will have $2020 in 5.88 billion, medical image analysis software system the size of the market will be $2.5 billion. Secondly, the AI + image system will be extended through the cloud, and the remote imaging diagnosis is in line with the fairness and efficiency of the state-sponsored medical treatment. The overall hope is to reduce the expenditure of medical expenses and help the health care control fee.
The foreseeable future
The most important role ai should play is the doctor's "assistant" role, not the protagonist of medical decisions. With its algorithms and databases for decision support, the decision maker should still be a doctor who can judge the actual situation.
The timely and accurate diagnosis of disease is the basis of clinical medicine, while the clinical diagnosis of western medicine relies on medical imaging. We believe that the increment of the future will focus on medical imaging software and online portal, integrated video diagnostic tools, computer aided diagnosis recognition, 3 d medical image, image quality control system.
From the domestic medical environment, the biggest challenge is the closure of medical data. Traditionally, every hospital has formed "information island". Of course, the situation has changed, for example, in Shanghai, medical data has been implemented in the shenlian system, and the number of hospitals participating in the shenlian system has increased to 40. With the rise of "patient-centered" medical service ideal, it is inevitable that medical imaging data flow smoothly between medical service units.
Currently, international brands account for more than 80 percent of the market for medical imaging equipment, with the top three GPS companies -- GE, philips, Siemens -- occupying more than 70 percent of the market. However, as the market continues to be developed, it is believed that there will be native excellent technology and the living space of commercial projects in the subdivision area.
Finally, the application of artificial intelligence technology in medical imaging diagnosis, was still in the stage of relatively weak ai, the model of image algorithm is constructed and deep learning can complete assigned tasks of doctors, auxiliary diagnosis, can improve reading efficiency, improve the diagnostic accuracy rate, but how to commercialization? Still a challenge! The key is who will pay?
I look forward to early intervention of commercial medical insurance, promote the development of local AI + images quickly, improve the efficiency and quality of medical behavior at the same time, reduce the misdiagnosis rate and complications of patients, thereby reducing the overall spending on health care costs, artificial intelligence technology in the image on the direction of medical applications will be taken from reduce medical expense of the corresponding commercial return. (wang Chen)
About the author:
Wang Chen, founding partner of jingxu, is responsible for equity investment in medical devices and medical services.
In more than 20 years of his career, Mr Wang Chen successively at philips medical treatment in the strategy of mergers and acquisitions and strategic consulting, as senior product manager at GE healthcare, and Johnson &johnson Cordis heart and peripheral vascular interventional treatment department, GSK prescription drugs division, roche tumor department, German boehringer ingelheim pharmaceutical core position. In addition, Mr. Wang Chen was a radiology resident. He graduated from the department of clinical medicine of Shanghai jiaotong university. She is currently a master of business administration at the university of wales.
|
| |
Previous article：Medical service level significantly improved (five years of development)
Next article：Why do hospitals buy dynamic DR?
|
| |
|
|
