Ever threatened to lay off doctors and lose more than one billion years in a row. Medical artificial intelligence is ready to abandon Beijing, Shanghai and Guangzhou?

"I can't do it, I can't win!"

Five years ago, after Ke Jie was defeated 3-0 by AlphaGo, she lost control of her emotions and trembled
.

This battle between human talent and sophisticated algorithms has changed the human perception of artificial intelligence
.

In the face of a "perfect" and "calm" opponent, even a top player like Ke Jie can only bow down

Artificial intelligence is endowed with various possibilities, and at the same time it brings unlimited reveries about whether it can eventually replace humans
.

The intervention of capital has accelerated the pace of artificial intelligence
.

Today, five years later, this technology has spread across all walks of life

However, for the future of artificial intelligence, the controversy and optimistic voices seem to have never stopped
.

AI medical care eager for cash flow

AI medical care eager for cash flow

With the sound of the Hong Kong Stock Exchange, the dust settled on the “first share of medical AI”
.

At the beginning of November 2021, Eagle Eye Technology officially landed on the Hong Kong stock market
.

The company is the first company of its kind to obtain the third-class medical device certificate of the State Food and Drug Administration, and its products are approved to assist in the diagnosis of diabetic retinopathy

Prior to this, many companies have been involved in the battle for the “first share of medical AI”
.

Since the beginning of this year, in addition to Eagle Eye Technology, many AI medical companies such as Keya Medical, Predictive Medical, and Shukun Technology have submitted prospectuses to the Hong Kong Stock Exchange
.

In addition, companies such as Huiyi Huiying and Shenrui Medical also have the potential to go public

In this crowded "golden track", it is no easy task to land on the beach with one or even several positions ahead of competitors
.

Keya Medical was the first medical AI company to submit an application to the Hong Kong Stock Exchange, but the listing application is now in an invalid state
.

Earlier, the former star company Yitu Technology had become the "acquired" party after its sprint IPO failed

The deeper reason behind the listing of AI medical companies is their desire for cash flow
.

Data shows that from 2019 to the first half of 2021, Sukun Technology has accumulated losses of more than 300 million yuan; although it is assumed that the loss of medical care in the first quarter of this year has shrunk to 17.
37 million yuan, in 2019 and 2020, its losses It has accumulated nearly 1 billion yuan
.

From the prospectus of Yingpu Technology, the company lost 87.
139 million yuan, 79.
626 million yuan and 37.
49 million yuan in 2019, 2020 and the first half of 2021, respectively
.

Rapid revenue growth and huge losses have almost become an imprint that the AI ​​medical industry cannot erase
.

However, the listing of Eagle Eye Technology also brings a glimmer of hope to the entire AI medical industry

Maintaining high barriers and heavy research and development drag down performance

Maintaining high barriers and heavy research and development drag down performance

As a "new technology" combined with practical applications, artificial intelligence has not really exploded for a long time.
Especially for the medical industry, AI companies still need to improve the accuracy of algorithms through continuous learning and identifying a large number of cases
.

Therefore, maintaining ultra-high R&D efforts is a hurdle that AI companies cannot overcome, and this is one of the fundamental reasons why the industry cannot be profitable
.

An AI medical company executive told the health industry that labor is the biggest cost in the research and development process
.

According to Yingpu Technology's prospectus, the company's R&D team has more than 80 members.
From 2019 to the first half of 2021, the total R&D expenditure has exceeded 100 million yuan, accounting for half of operating costs for a long time
.

In the first half of 2021, Sukun Technology's R&D expenditure reached 81.
18 million, accounting for 154% of its operating income in the same period
.

Take Yingpu Technology as an example.
At present, its products can identify 55 health risks such as diabetes, hypertension, anemia, and arteriosclerosis
.

But in fact, there are more than 200 common diseases that can be identified through the retina

For AI medical companies, perhaps in the next few years, they will not be able to get rid of the pressure of R&D expenditure
.

Re-research and development comes from the urgent need to improve product performance on the one hand, and on the other hand to avoid being overtaken by opponents
.

Judging from the AI ​​products that have obtained the three types of medical device certificates, the products of many companies have overlapped in function

Sukun Technology stated in the prospectus that the company needs to invest a lot of resources in research and development to lead technological progress, so that its products are innovative and competitive in the market
.
It is expected that in the future, the company's research and development expenditure will continue to increase
.

Ma Chune, one of the founders of Shukun Technology, also publicly stated that compared with other AI products, choosing cardiovascular and cerebrovascular diseases as the entry point is the increasing demand for doctors and patients on the one hand, and the complicated diagnosis process of cardiovascular diseases on the other hand.
, The technical threshold is high, and many primary hospitals are unable to develop it
.

It can be seen that R&D strength is the key to maintaining the company's high barriers in segmented fields
.

Whether it can replace the doctor is controversial

Whether it can replace the doctor is controversial

The continuous increase in R&D investment has enabled AI technology to achieve leaps and bounds in the practical application of the medical industry
.

For example, the performance of the CheXNet model developed by Stanford University exceeds the average performance of radiologists in pneumonia examinations
.

Through a large amount of data comparison, it takes about 10 minutes for a director-level expert to view hundreds of CT images continuously, while AI only takes less than 1 minute
.
In the view link, AI can greatly improve efficiency and free up more time for doctors for academic research
.

For a long time, due to the super learning ability of AI technology, people were worried that the technology will eventually replace humans
.

However, Zhang Dalei, founder of Eagle Eye Technology, pointed out that many people think that artificial intelligence may replace doctors, but the history of more than five thousand years has proved that the emergence of all technologies will eventually become tools that doctors can use
.

Many people think that artificial intelligence may replace doctors, but the history of more than five thousand years has proved that in the end all technologies will eventually become tools that doctors can use
.

Unlike AI practitioners who are consistently optimistic about the prospects of the industry, there seems to be some controversy over AI products in the circle of doctors
.

Liu Yang, deputy chief physician of the Department of Orthopedics at Beijing Chaoyang Hospital, told the health community that although AI technology will play a certain auxiliary role in practical applications, for experienced doctors, there is still a lot of room for improvement in AI at the technical level
.
Compared with experienced surgeons, doctors with relatively lack of surgical experience are more likely to accept AI products, which can help them improve surgical safety
.

.
Compared with experienced surgeons, doctors with relatively lack of surgical experience are more likely to accept AI products, which can help them improve surgical safety
.

Liu Yang believes that AI has demonstrated excellence in surgical procedures, especially in joint prosthesis replacement operations, which can improve the accuracy of surgical osteotomy and more accurate implantation of joint prostheses; in addition, for example, in spine surgery vertebral arch During the nail implantation process, AI assistance will also make the path of the screw placement more accurate, which can greatly improve the safety of the operation
.
However, the current AI-assisted products are still inadequate.
In practical applications, repeated positioning is required, which increases the operation time
.

"AI is the future development direction is unquestionable, and it is also the general trend.
It can assist surgery more accurately and improve the safety of surgery.
In the long term, AI technology may replace the role of doctors in specific scenarios
.
" But Liu Yang also said.
"But in the short term, AI cannot completely replace manual labor, especially for preoperative planning.
Because the specific conditions of each patient are different, it is difficult for AI to make a corresponding surgical plan
.
"

Is the three types of certificates a cornucopia?

Is the three types of certificates a cornucopia?

The ability of AI medical products to enter hospitals and be inspected by doctors has a direct relationship with the company's successive approval of three types of medical device certificates starting in 2020
.

Only companies that have obtained the three types of certificates have the prerequisite for commercialization
.

Only companies that have obtained the three types of certificates have the prerequisite for commercialization
.

The new version of the "Medical Device Classification Catalog" clearly stipulates that if the AI ​​diagnostic software automatically recognizes the lesions through algorithms and provides clear diagnostic prompts, it must be managed in accordance with the third category of medical devices
.

If the AI ​​diagnosis software automatically recognizes the diseased location through algorithms and provides clear diagnosis prompts, it needs to be managed in accordance with the third category of medical equipment
.

From the effect point of view, AI companies have indeed gained new life from it
.

Data show that Yingpu Technology's revenue in 2019, 2020 and the first half of 2021 will be 30.
415 million, 47.
672 million and 49.
477 million respectively
.
From the data point of view, the income in the first half of this year has exceeded that of the entire year of last year
.
In 2020 and the first half of 2021, Sukun Technology achieved revenues of 24.
71 million and 52.
624 million respectively.
The revenue in the first half of this year has more than doubled that of last year
.

However, although AI companies have built a higher barrier in the industry by taking "certificates", they will still encounter various problems in actual promotion
.

First of all, it takes a long time to obtain a certificate, which usually takes about three years
.
Secondly, in the process of selling to hospitals, hospital procurement often requires a long process, and how to set prices still needs to be approved by the local price bureau
.

The above-mentioned AI corporate executives admitted frankly that in addition to the hospital procurement process and price bureau pricing, as a new thing, AI will also break the existing pattern to a certain extent
.
Therefore, there are bound to be limitations in individual hospitals that exclude psychology and cognition
.

As a new thing, AI will also break the existing pattern to a certain extent
.

Sinking to third- and fourth-tier cities

Sinking to third- and fourth-tier cities

As of the close of November 23, Eagle Eye Technology reported HK$63.
95 per share, which was down from the issue price
.
This unexpected performance aroused market concerns about the commercialization prospects of AI companies
.

As of the close of November 23, Eagle Eye Technology reported HK$63.
95 per share, which was down from the issue price
.

However, according to Frost & Sullivan’s prediction, China’s artificial intelligence medical impact market is expected to grow from 300 million yuan in 2020 to 92.
3 billion yuan in 2030, with a compound annual growth rate of more than 76% during the period
.

In fact, in the face of such a big cake, AI medical companies have already planned
.
In the prospectus, Yingpu Technology has basically outlined a clear commercialization path
.

At present, Yingtong Technology has obtained pricing guidelines from five provinces including Hebei, Shandong, and Shanxi
.
According to the guidelines, hospitals can charge patients individually for the provision of medical services, and at the same time can better evaluate the return on investment
.
The prospectus shows that the company has reached initial purchase intentions with 200 hospitals
.
In addition to the resources in the hospital, Yingtong Technology will also shift its focus to the field of health, and expand to more scenarios such as insurance companies, eyewear companies, and pharmacies
.

According to the report of Yiou Think Tank, the hospital-end paid penetration rate of AI medical imaging is only 4.
5% to 7% at this stage
.
As the value of the product is recognized, hospitals will increase their willingness to pay, and the penetration rate is expected to reach 15% to 20% in 2023
.

As the value of the product is recognized, hospitals will increase their willingness to pay, and the penetration rate is expected to reach 15% to 20% in 2023
.

In this regard, Essence Securities pointed out in the research report that for the business model, AI medical imaging products will enter hospitals in the form of medical devices
.
In the long run, with the further improvement of product functions and the continuous penetration of the primary medical market, the industry will also explore a model of sharing medical service fees, that is, charging according to the amount of diagnosis or the amount of calls
.

In China’s remote areas, many cities lack expert resources, causing some patients to turn to Beijing, Shanghai, Guangzhou and other large cities for medical treatment.
This has also aggravated the congestion of medical resources, and the waiting time in the queue may be too long, which may delay the condition
.

The head of the marketing department of an AI company told the health industry that in the future, the company will shift its focus to third- and fourth-tier cities.
These areas have greater demand for AI products and even plan to sink to county-level hospitals in the future
.
Empowering primary care with AI is the best commercialization path in the future
.

"AI is a new thing, and there must be a process of acceptance
.
Doctors in different regions and different hospitals are at different levels, and there are also deviations in human judgments at a certain point.
AI can just solve these problems
.
" The person in charge predicts , "When AI is popularized to a certain extent, there may be an arms race in the future, and patients will be inclined to hospitals equipped with AI technology
.
"