Opportunities and challenges of artificial intelligence entering the medical field

Since the concept of artificial intelligence (AI) has become popular, its entry into various industrial fields has accelerated. However, given the complexity of the medical field, the expansion of AI in the medical industry has not been smooth. The only one that has achieved large-scale positive growth is surgical robots. After many years of development, the entire field is still an immature early market. It is difficult to truly commercialize.

If divided according to the fields in which artificial intelligence enters the medical field, the main categories are drug research and development, diagnosis and treatment, image recognition, surgical robots and health management. However, from a practical point of view, although there has been some progress in the field of drug research and development, there is still a certain distance from true scale. The tool attributes in the field of diagnosis and treatment are stronger, and their role in medical treatment only plays an information-based role. The areas that have really attracted the market in the past five years are AI solutions (mainly image recognition), surgical robots and health management.

Although the capital market is enthusiastic about these fields, from the current point of view, the real development prospects are surgical robots. Without strategic transformation, the other two fields will be difficult to develop commercially even in the long term. Scale.

Four angles of analysis

These three can be analyzed from four angles: regulatory attributes, practicality, urgency and sustainability.

First of all, from the perspective of regulatory attributes, regulatory licenses are obtained based on medical devices. However, only surgical robots truly rely on medical devices and consumables to navigate the business model. Others are regulated based on medical devices. License, but in fact it does not rely on the medical device itself for development, but prefers the in-hospital and out-of-hospital model of medical informatization. Image recognition is similar to in-hospital imaging departments purchasing systems to improve their information capabilities, while digital therapy is similar to using information software to carry out specialized out-of-hospital follow-up and rehabilitation management. From the perspective of regulatory access, the medical device and consumables model is a mature business model with a high unit price per customer; while the specialist information software model has a lower unit price per customer, and the growth of the market scale depends on volume. As for health management outside the hospital, even the logic of the business model cannot be established.

Secondly, from a practical point of view, surgical robots and image recognition are highly practical, and the users are mainly hospital departments with relatively clear purchasing needs. The efficacy of digital therapy has only been proven in trials, but since it has never gained a large base of users, its practicality is questionable. Since it requires a prescription from a doctor and is mainly targeted at out-of-hospital C-end customers, the use scenario of digital therapy cannot create an urgent need scenario like the first two. Everything is controlled by the subjective will of the individual, and the sustainability is weak.

Thirdly, urgency determines the ability and willingness to pay. Low urgency will lead to low willingness to pay, and vice versa. From the perspective of willingness to pay, the urgency of surgical robots is high and the willingness to pay is high, but the urgency of image recognition is not strong, and the urgency of digital therapy is even weaker and the willingness to pay is lower. From the perspective of affordability, hospitals have the most sufficient funds to purchase equipment and the highest market acceptance. The challenge facing the sales of surgical robots is policy access rather than affordability. Image recognition is limited to imaging departments, and funding is relatively limited, so the ability to pay is obviously limited. Moreover, the services provided by image recognition are more similar to information software, making it difficult to increase the unit price per customer. As for digital therapies, the ability to pay out-of-pocket on the consumer side is always a big problem. It is difficult to promote the commercialization of the industry and must rely on payers. However, payers need to see clear efficacy and cost control. Digital therapies do not have this, and it is difficult to obtain commercial benefits. the required scale.

Finally, from the perspective of sustainability, the key is whether the business model can continue to meet customer needs and payment capabilities. Surgical robots and image recognition mainly rely on hospital procurement, and have strong sustainability capabilities, but digital therapy relies on The C-end is self-funded, making it difficult to obtain a stable market scale and having weak sustainability.

Challenges of scale

In essence, image recognition helps doctors more in terms of efficiency, but unlike global information systems such as electronic medical records, medical institutions There is no strong urgency to improve local efficiency. Therefore, under the premise that the willingness to pay of medical institutions is not strong, the pricing of image recognition products can only be compared to specialized information systems, and it is difficult to compare with the information systems of large hospitals. In a market environment where demand is limited and pricing is difficult to improve, the commercialization of image recognition faces great limitations.

On the other hand, image recognition is a labor-intensive service and requires a large amount of manpower to produce biomarkers. This has greatly increased the company's costs. The cost of every 1 yuan of revenue earned is Far more than 1 yuan, such a business model can only be supported by continuous financing. Unlike traditional Internet companies, because it is a completely 2B market and the unit price per customer is low, image recognition companies lack the possibility of rapid scale-up, and it is difficult to burn money for long-term development.

Although the field of surgical robots is relatively narrow, the currently mature ones are mainly abdominal cavity, mainly urology and general surgery. The scale of orthopedics will still take time, but because it can solve the pain points of doctors, it can greatly improve the surgical operation. The accuracy and efficiency have achieved significant development after several years of market cultivation. Due to the high unit price per customer and a certain base of self-paying users, surgical robots will see significant growth as medical insurance is included in some areas.

However, the surgical robot market still lacks significant domestic brands. Although there are currently many product lines under development and clinical trials, there is still a lack of truly scalable surgical robot products. Surgical robots are a long-term development business model, but opening up sales channels in the hospital market must be closely coordinated with product development. Otherwise, it is easy to form a situation where there is a product but no market, and it will drag on for many years, unable to generate cash flow, and always need financing. blood transfusion.

The main manifestation of artificial intelligence in the field of health management is digital therapy. Digital therapy is not a new product, but more of a software-plus-hardware model that repackages existing products. Unlike image recognition and surgical robots, digital therapy is more individual-oriented. The essence of artificial intelligence is to improve efficiency and skills and reduce costs, but digital therapeutics do not reflect this. While digital therapeutics have demonstrated in trials that they can effectively treat and manage chronic or mental health conditions, in practice, their actual effectiveness remains questionable. As a result, medical insurance and commercial insurance are still very cautious as payers, and their willingness to directly cover digital therapies is very low, reducing the possibility of rapid scale-up.

Of course, the main advantage of artificial intelligence in digital therapy lies in its algorithm. It continuously adjusts user data to promote the algorithm to be more accurate, thereby improving the user's health level. However, just like image recognition, high manpower investment and large samples are required to develop algorithms with higher accuracy. The current digital therapy adopts the research and development model of drugs, and only conducts trials on small groups of people in hospitals. When such digital therapy products are launched, the user usage and renewal rates are not high. The actual prescription dispensing rate of Pear Therapeutics, as the leader, is only 50%, and the actual prescription payment rate is only 25%, which greatly restricts its potential to scale.

Therefore, if we judge from the potential of the business model, the urgency of user needs, willingness and ability to pay, and sustainability are the keys to judging whether it can achieve real development in the future. Judging from market trends, the business model of surgical robots has matured, but is subject to technological capabilities and future scale depends on technological maturity and cost control. The business model of AI solutions based on image recognition is still immature, and whether it can truly scale up in the future is subject to whether new rigidly needed application scenarios can be found and the hospital's willingness to pay increases. Digital therapy does not yet have a clear business model, and the market will continue to be explored. ????

The above is the detailed content of Opportunities and challenges of artificial intelligence entering the medical field. For more information, please follow other related articles on the PHP Chinese website!