The "Guidelines for the Establishment of Radiological Examination Price Projects (Trial)" recently released by the National Medical Insurance Administration clearly defines the business model of imaging AI, which has attracted widespread attention in the industry. The guideline aims to standardize the price of radiological examination items and support the promotion of artificial intelligence-assisted diagnosis in clinical applications while avoiding increasing the financial burden on patients. The editor of Downcodes will provide an in-depth interpretation of this policy and analyze its impact on the imaging AI industry.
On November 20, the National Medical Insurance Administration issued a "Guidelines for the Establishment of Radiological Examination Price Projects (Trial)", which not only integrated and standardized the current radiological examination projects, but also defined the business model of imaging AI.
The policy interpretation article pointed out: "Artificial intelligence technology plays a role in assisting diagnosis or improving efficiency to a certain extent in clinical practice, but it cannot yet replace physician diagnosis. In countries where there is no independent medical service output and the quality and effectiveness of auxiliary diagnosis are difficult to determine, Under such circumstances, after the examination fees for the corresponding diagnostic items have been collected, it is not appropriate to charge the patient additionally for artificial intelligence-assisted diagnosis alone.
In this regard, in order to support the clinical application of artificial intelligence-assisted diagnosis and prevent additional burden on patients, the project establishment guidelines uniformly arrange the expanded items of "artificial intelligence-assisted diagnosis" under the main radiological examination items. That is to say, hospitals that use artificial intelligence for auxiliary diagnosis will charge the same price as the main item, but will not be charged twice as much as the main item. "
To put it simply, the National Medical Insurance Administration supports the widespread application of imaging AI in clinical practice, but does not allow hospitals to pass on the costs of using AI to patients.
Faced with the new policy, imaging AI practitioners have mixed feelings. Fortunately, the National Medical Insurance Bureau officially recognized the contribution of AI to clinical practice and let relevant parties know about the use of AI. But I am also worried: when the introduction of AI cannot bring direct benefits to the hospital, will the given new policy be enough to support the revenue of the imaging AI industry?
In the early stages of the development of imaging AI, the route set by start-up companies for medical AI is to expect it to pass through market access, price access, and medical insurance access one by one to form independent medical device products , and ultimately achieve regular payment to patients, creating a closed loop. The solution is implanted in the hospital.
There are precedents for such a business model. For example, Digital Diagnostics in the United States analyzes diabetic retinopathy at US$55 per time (2022 data, the same below), and Viz.AI US$1,040 per large blood vessel blockage test. They are all domestic imaging AIs. The company's compass in its early years.
Following this path, companies such as Keya Medical and Eagle Eye Technology began to promote price access and medical insurance access on a large scale after their products obtained Class III certificates. Over the past few years, related products have successfully entered provincial price lists in more than ten provinces and cities, theoretically making it possible for patients to pay out of pocket. However, they have encountered challenges in the more important aspect of medical insurance, and have only entered basic medical care in a few areas. The scope of insurance payment is far from being large-scale.
There are multiple reasons for the failure of this path. In the past few years, enterprises, governments, and regulatory agencies have jointly promoted hospital prices and payment models, but overall they are not active enough.
On the one hand, large-scale price access and payment model verification in medical insurance access require companies to spend a lot of manpower and material resources to implement, but the results achieved cannot guarantee that the products will achieve considerable commercialization results, which limits the speed of advancement.
On the other hand, price access and medical insurance access are similar to public goods, and there is the possibility of investment in the former and free riding on the latter. Therefore, companies that are the first to invest in related research tend to hide phased research results, resulting in a limitation of the overall advancement speed of the industry. , and easily lead to repeated research on a single product.
Nowadays, the introduction of the new policy has undoubtedly shattered the dream of imaging AI as an independent product to seek regular payment by medical insurance. Sustainable business models that are common to drugs and devices may never be implemented in the field of imaging AI.
Although an important theoretical path to commercial realization has been lost, the New Deal has not had a great negative impact on the profitability of imaging AI companies. Instead, it has pointed out the direction for the long-term development of imaging AI companies.
Let’s talk about the impact of policy first. At this stage, the revenue of imaging AI companies has very little to do with medical insurance. They mainly rely on bidding to sell imaging AI to hospitals in a buyout or SaaS model. In addition, medical imaging has always been an important source of paper output. Many hospitals and doctors are willing to seek cooperation with AI companies to improve the output quantity and quality of relevant scientific research results.
In addition, cooperating with imaging equipment manufacturers and paying directly to the equipment manufacturers is also an important way for medical AI companies to make profits. This is a win-win cooperation. Imaging equipment companies can quickly obtain a large number of applications through intelligent algorithm licensing, effectively improving the competitiveness of their own products. Hospitals also prefer to directly call algorithms from the imaging equipment manufacturers' platforms to increase the revenue of imaging AI companies. In the early years, United Imaging Group specifically established United Imaging Intelligence to overcome imaging problems in various scenarios, and it has become one of the largest imaging AI companies with the most complete products. Later, MNCs such as GE Healthcare and Philips Healthcare also established AI ecosystems in China, including A large number of high-quality partners.
With these diversified methods, imaging AI companies have implemented imaging AI in a large number of hospitals and achieved hundreds of millions in revenue without the support of medical insurance.
The main sources of income for imaging AI companies
Let’s talk about policy guidance for imaging AI. The article mentioned "supporting the clinical application of artificial intelligence-assisted diagnosis", which is actually an affirmation of the clinical application of artificial intelligence. In practice, some hospitals in my country have carried out separate projects for services involving imaging AI. After the hospital uses imaging AI for auxiliary diagnosis, a part of the revenue can be allocated as a reward for the performance of the AI to provide services to imaging AI companies.
For example, Shandong Province has done a lot of innovative work in AI charging design. Some hospitals use AI to perform early cancer screening CT scans. The actual price is 340 yuan per part, which includes 50 yuan for artificial intelligence screening assistance. Diagnosis costs (medical insurance does not participate in payment).
However, it should also be noted that it is difficult for imaging AI companies to feel the benefits brought by the introduction of the new policy in the short term. Currently, the number of hospitals that use performance allocation to pay for AI is sparse, and the inspection items covered are also quite limited. Therefore, it may take several years for small-scale trials to be implemented on a large scale, and more refined policies are needed to promote the establishment of a new payment system.
In addition, after the payment path for medical insurance cases is blocked, imaging AI companies will rely more on hospitals and imaging equipment companies as payers. Due to the high-pressure medical anti-corruption, the total amount of hospital medical equipment procurement in the first half of 2024 was almost halved, and the winning bid amount for magnetic resonance imaging and CT was only 60% of the same period last year. Under this situation, the pressure faced by imaging equipment manufacturers in the upstream of the industry chain will be directly transmitted to imaging AI companies in the midstream, and the latter's revenue will suffer a certain scale of decline before the demand for equipment procurement is released.
At the end of the article by the National Medical Insurance Administration, the article pointed out: “The project establishment guide uniformly arranges the expanded item of “artificial intelligence-assisted diagnosis” under the main item of radiological examination in order to reflect the functional positioning of artificial intelligence technology in improving quality and efficiency. rather than increasing costs. ”
This sentence not only applies to imaging AI, but may also apply to various artificial intelligences in the medical industry.
The "Notice on Regulating the Use and Charges of Surgical Robot-Assisted Operating Systems" issued by the Hunan Medical Insurance Bureau in 2022 has unified the form and price of surgical robots in the form of policy guidance. Its essence is to ensure reasonable medical insurance expenditures and reasonable patients. Guide the orderly development of relevant markets under the premise of expenditure , and prevent enterprises and hospitals from "innovating" in charging items with the help of simple software.
Nowadays, the introduction of the "Guidelines for the Establishment of Price Projects for Radiological Examinations (Trial)" has the same purpose. It establishes the positioning of imaging AI. It is hoped that AI can help hospitals improve quality and efficiency, and bring increment to the entire medical system. And then reflect your own value.
Combining the two policies, it is not difficult to find that the formulators do not support companies using AI as an independent product or a selling point of independent products. Instead, they hope that it can be used as a tool for equipment and systems, like automatic navigation in the automotive industry or quality control in the industrial industry. part to support its greater value.
In reality, the so-called "head imaging AI companies" have long torn off the label of "imaging AI", designed highly intelligent hardware or systems, and become complete medical device companies or medical IT companies.
Shenrui Medical has made many achievements in the field of medical IT. After the rise of big models, the company focused on hospital data management and built a multi-modal data management engine covering the entire process of data collection, management, and data labeling, as well as a multi-modal large language model, an image general model, and a multi-modal large model. Modal AI engine; and provides multiple capability opening models in multiple forms such as full-cycle governance capability opening, data service customization capability opening, and multi-modal AI modeling capability opening.
In addition, for the data asset solutions that are urgently needed by hospitals, Shenrui Medical also integrates AI to provide smart management, smart scientific research, smart clinical, AI innovation center and other scenarios for medical institutions to provide asset management-related smart products and services. Serve.
In terms of medical devices, Shukun Technology and Infer Medical are making plans. Relying on AI, Shukun Technology has independently developed native ultrasound hardware equipment such as "Turing Brain" and "Turing AR", enabling it to deeply integrate intelligent algorithms. It can not only collect all organ information during ultrasound diagnosis and treatment, but also reflect lesions in real time. It also optimizes the doctor’s experience and prevents doctors from using the “second screen” in clinical practice.
In the view of Shukun Technology, the integrated collaboration of software and hardware is the reconstruction of user experience, and it is also the key for AI to continue to break through the boundaries of technology. In the future, every piece of hardware will move from the industrial era to the AI era, and every step and every second of the interaction between doctors and equipment will be supported by AI.
Infer Medical has entered the surgical robot track and deeply integrated imaging AI into the hardware. For example, the company's self-developed AI navigation robot "Longdianjing® Puncture Surgery Robot" adds AI intelligent technical support based on magnetic navigation guidance. Empowered by intelligent algorithms, it is speculated that it can achieve fully automatic tissue lesion identification and reconstruction, and further perform automatic surgical path planning, puncture guidance, and post-ablation evaluation, thereby effectively assisting doctors to complete percutaneous puncture operations more accurately and quickly.
At this point, all former imaging AI companies have completed their own value reconstruction. The end of the old era corresponds to the advent of a new era for AI companies.
The new policy of the National Medical Insurance Administration has pointed out the direction for the development of the imaging AI industry. Although it may face some challenges in the short term, in the long term, imaging AI companies still have broad prospects by cooperating with hospitals and imaging equipment manufacturers and transforming their own business models. development prospects. In the future, imaging AI will no longer be an independent product, but will be integrated into medical equipment and systems to improve medical efficiency and quality.