Zheng Lingxiao of Ruixing Medical: The Medical AI Industry Should Not Follow a 'Software Sales' Model — Pay-Per-Use Is the Future

By Yu ShiqiThe AI medical imaging sector has arguably been the most closely watched, yet also the most volatile, track over the past five years. The initial "compelling narrative" pitched by the entire industry was that AI technology would disruptively enhance the efficiency of clinical imaging diagnosis. The capital market responded enthusiastically, with funding surging to a peak in 2017 when the sector's total annual financing exceeded RMB 10 billion. Tech giants including Tencent, Alibaba, and Baidu all entered the market. As one industry insider put it, at its height, there were as many as 200 companies competing in the highly sought-after lung nodule detection segment.

However, the prolonged inability to achieve commercialization has repeatedly poured cold water on the entire sector. For medical AI enterprises, this represents an emerging domain, and the regulatory review and approval process for their products is exceedingly complex. Without Class III medical device certification, products can only be provided to hospitals for free use and cannot be included in hospital billing catalogs, naturally preventing the establishment of a viable commercial closed loop. As the tide recedes, the two hundred companies developing lung nodule AI products have quietly faded from view. Even the domestic leader in medical AI imaging, Yitu Healthcare, has exited the market, selling its medical AI business to Deepwise.

Last year, the industry experienced a notable turning point. In January 2020, the National Medical Products Administration (NMPA) approved the first Class III medical AI device. Over the subsequent year or so, more than a dozen products obtained regulatory approval. Since then, the significant tailwinds generated by these product approvals have rapidly revitalized the sector, making it once again a major investment hotspot. Established venture capital firms, including Sequoia China and IDG Capital, have successively stepped in to invest.

Following the initial opening of the regulatory certification pathway, commercialization has emerged as the core priority for the first batch of approved companies. The overall market remains in a very early stage. According to data from a Frost & Sullivan report, the market size for AI medical devices in China was only RMB 300 million in 2020, with a projected figure of approximately RMB 500 million for 2021. However, the sector is poised to enter a period of explosive growth thereafter. By 2025, the market size is expected to reach RMB 20.4 billion, representing a compound annual growth rate (CAGR) of 134% from 2020 to 2025.

RaysightMed stands out as a leading company among this wave of enterprises that have secured regulatory approvals. Founded in 2017, the company obtained the Class III medical device registration certificate for its inaugural medical AI product in April this year, enabling its market launch. In August this year, it successfully closed its Series C financing round, backed by prominent venture capital firms including Matrix Partners China, Tencent, and SDIC Capital.

We spoke with the company's CEO, Zheng Lingxiao, who holds a highly pragmatic understanding of the industry's evolution over recent years, as well as the market prospects for the medical AI imaging sector in China. Naturally, they have also embarked on their commercialization efforts, and he discussed at length the challenges encountered during the process.

Economic Observer Healthcare: What is your current main product, and how does it differ from conventional approaches?

Zheng Lingxiao:Our first commercially available product is an FFR-CT (coronary blood supply function assessment software), which primarily replaces invasive coronary angiography.

Currently, invasive coronary angiography primarily serves two purposes: diagnosis and treatment. The diagnostic component assesses the degree of stenosis in the patient's coronary arteries, which serves as the criterion for determining whether stent implantation is necessary.

Traditional coronary angiography is an invasive procedure that requires local anesthesia and involves interventional techniques, during which one or more catheters must be inserted into the blood vessels. However, current international statistics indicate that approximately 70% of patients undergoing coronary angiography do not actually require stent implantation. In China, based on our communications with numerous hospitals, this proportion is approximately 60%.

We believe this process should actually be replaced by a non-invasive method, namely FFR-CT. Patients would follow the same workflow as before by simply undergoing a CT scan, and our FFR-CT product can then analyze the CT images, eliminating the need for mandatory invasive coronary angiography.

Economic Observer Health: This sounds like it changes the entire workflow. How do doctors and hospitals perceive this?

Zheng Lingxiao:From an economic perspective, the cost analysis is straightforward. Currently, a coronary angiography procedure costs approximately 10,000 RMB. Even with medical insurance covering 70%, patients' out-of-pocket expenses still amount to around 3,000 RMB. The FFR-CT product is priced between 2,000 and 3,000 RMB. This eliminates the need for patients to undergo additional invasive procedures, effectively bringing the overall cost down to the same level as before.

These cover only the direct medical costs. Many patients from other regions travel to hospitals in major cities to undergo this procedure, incurring additional expenses for travel and accommodation. If FFR-CT were available locally, they could have their medical needs addressed directly.

What is the driving force for hospitals? If we look at the revenue data, a substantial portion indeed comes from surgical departments, such as the Department of Cardiology. However, when assessing actual profitability, these departments may yield zero profit or even operate at a loss. In reality, hospital profits primarily stem from clinical laboratory and medical imaging departments. Therefore, performing a coronary angiography procedure is actually unprofitable and merely results in the inefficient utilization of hospital beds.

Increasing the clinical adoption of FFR-CT can directly enhance physicians' diagnostic capabilities and accuracy. In the past, a physician might perform coronary angiography procedures from early morning until late at night, or even into the early hours, yet only a small fraction of these cases actually required subsequent intervention. This has essentially resulted in a waste of valuable hospital bed resources while yielding limited financial returns for both physicians and hospitals.

Strictly in terms of market size, there are currently millions of patients in China each year who are suitable for FFR-CT analysis. At a per-case cost of 2,000 to 3,000 RMB, this represents a potential market in the tens of billions of RMB.

Economic Observer Healthcare: If the market opportunity exists, why are there so few companies in China's medical AI industry focusing on this direction? Currently, we only see two companies: yours and Keya Medical?

Lingxiao Zheng:The primary reason is that in this field, both we and Keya Medical have advanced at a relatively fast pace, and the industry has likely already recognized our clear leading advantage. However, three or four years ago, there were actually around ten companies of varying sizes operating in this space, with major enterprises like United Imaging Intelligence and Lepu Medical also having established their strategic presence.

This field is significantly more complex than others. At the time, I had extensive discussions with my two other co-founders. Back then, Ma Jun was working on the da Vinci Surgical System; in fact, his expertise was highly aligned with the demands of this AI wave. Consequently, companies like Alibaba, Tencent, and numerous newly established startups all approached him. However, at that time, the industry's primary focus was on developing products for pulmonary nodule analysis.

We believe that the technological barriers for lung nodule detection are not particularly high, which is further evidenced by the subsequent entry of numerous companies into this field. Frankly speaking, while many claim to be developing AI, they are essentially just modifying frameworks and models based on existing systems and then deploying them for application. This is precisely how many lung nodule products are actually developed.

Our primary consideration in selecting FFR-CT was its high technical barrier. Designed to serve as a "gatekeeper" prior to interventional procedures, it demands exceptional precision, with image resolution required to reach the level of a single pixel or even half a pixel. Since subsequent computational simulations rely on accurately reconstructing the entire coronary artery system, insufficient resolution would inevitably lead to imprecise lumen segmentation, thereby compromising the accuracy of the final results.

Another reason is the widespread initial skepticism toward the pay-per-use service model, with doubts about its viability as the industry remained accustomed to the traditional path of direct software sales. I recall a company developing similar products that only realized such technology actually required regulatory certification after Keya Medical obtained its certificate (In 2020, Keya Medical secured China's first Class III medical device registration certificate for AI, approved by the National Medical Products Administration). Previously, these were simply viewed as computer software development. It was only after our two companies obtained the certifications and were listed in the pricing catalogs (Medical devices must apply for a pricing code from local government authorities; only upon approval can they be added to hospital fee schedules to charge patients) that the pay-per-use service model was truly implemented.

EO Health: That’s a very interesting question. Why did you initially focus your revenue model on the patient side rather than the conventional hospital side?

Zheng Lingxiao:At that time, many medical AI companies primarily offered software products, but in China, selling software and servers is actually extremely difficult. I was working at Dassault Systèmes, an engineering software company, at the time, and I felt this reality acutely. When it comes to what China's manufacturing industry lacks, people might point to a shortage of high-precision, cutting-edge machining equipment. However, in reality, the most critical shortfall in China's manufacturing sector is engineering software—an area that remains essentially blank in China.

However, this was quite common overseas. At that time, a simulation software from Dassault Systèmes charged nearly $10 million annually in licensing or leasing fees to companies like Ford, General Motors, and Boeing. In China, there was simply no such market.

It is generally perceived merely as software, and subconsciously, it is not regarded as a particularly critical priority, resulting in a weak willingness to pay. Moreover, charging hospitals for software fees is even more challenging, as their payment willingness at this stage remains relatively low. The approximate market value for all the tools or software required by an entire radiology department in a single hospital is only around RMB 1 million. If a single software package is priced at RMB 500,000, and a hospital may need to purchase four, how many hospitals in China actually possess both the willingness and the financial capacity to make such a purchase?

Therefore, FFR-CT was conceptualized from the outset primarily as a diagnostic testing service, with costs borne directly by patients. Overseas, companies such as HeartFlow (a U.S.-based AI diagnostic firm specializing in coronary artery disease) have already emerged. They have successfully validated a pay-per-test commercialization model billed to patients and have also been integrated into commercial health insurance systems. From my perspective, the market size for FFR-CT as a standalone product alone exceeds that of the software sales model previously discussed.

Economic Observer Health: From the patient's perspective, what does the current end-to-end service payment process look like?

Lingxiao Zheng:In provinces where it has already been included in the official medical service pricing catalog, if we have also secured hospital listing approval, physicians can directly order the examination for patients.

In provinces where the service has not yet been included in the official medical pricing catalog, we currently collaborate with third-party imaging centers. One model involves patients undergoing imaging scans at these centers and additionally purchasing our analysis service. The other model allows patients who have already undergone a CT scan to simply upload their imaging data directly for FFR-CT analysis.

Economic Observer Big Health: This year marks the first year of commercialization. What are your current strategic plans?

Zheng Lingxiao:Following the regulatory approval secured in April this year, the marketing team has experienced explosive growth. All team members come from a healthcare background, with the vast majority specializing in the cardiovascular and cerebrovascular fields.

They are extensively engaging with both established clients and our carefully selected target customers to introduce our products. This is far more complex than simple domestic substitution, as we are pioneering an entirely new field. Advancing in a novel domain requires a high degree of professional expertise; you must be able to clearly articulate to experts exactly what specific benefits the product can deliver to their practice.

Once a relationship of trust has been established, we should not hastily push for product purchases. Instead, we invite experts to conduct trials, enabling them to recommend suitable and clinically meaningful cases for a free trial of our services. To date, we have successfully implemented these trials across 500 hospitals.

The Economic Observer Health: At what point will the trial transition to a paid service?

Zheng Lingxiao:Payment will only commence once we secure hospital access, which entails formal integration into the hospital's billing system. The entire access process is highly complex, requiring extensive coordination and communication with multiple departments, including Cardiology, Imaging, Information Technology, Procurement, and Medical Insurance. Currently, fewer than 50 hospitals are already progressing through this workflow, which is inherently a time-intensive endeavor.

Each stakeholder has different areas of focus. For instance, the IT department prioritizes how to ensure full regulatory compliance when hospital data is exported outside the facility, while the procurement department might question why a software license isn't sold outright instead of a pay-per-use service model. Ultimately, every department has its own specific requirements.

Therefore, I have always emphasized internally that going from 0 to 1 is the most challenging phase. We have consistently advocated for rolling out an implementable, replicable, and rapidly scalable model, which would make subsequent expansion much easier. However, establishing a standardized regional template is exceptionally difficult.

Economic Observer Healthcare: From this perspective, wouldn't partnering with third-party imaging centers be a more straightforward approach? Could this become the primary promotion strategy?

Zheng Lingxiao:This is indeed a straightforward path. However, diagnostic testing for coronary heart disease (CHD) patients currently remains concentrated in public hospitals, with relatively few patients utilizing third-party facilities. Therefore, I anticipate that public hospitals will undoubtedly serve as the primary market in the initial phase. Subsequently, a supplementary, comprehensive multi-channel sales model will incorporate third-party imaging centers, health examination centers, and internet hospitals.

Economic Observer Healthcare: Securing hospital market access independently is highly complex. Have you considered partnering with CT hardware manufacturers, given that their distribution channels are more established?

Zheng Lingxiao:Several major CT manufacturers are currently in discussions with us, all expressing interest in pursuing deeper collaboration. Due to the ongoing volume-based procurement (VBP) policy and the resulting market saturation, CT manufacturers need to develop new business segments with growth potential.

Based on our own assessment, the market value of the entire intelligent diagnostics sector should be comparable to that of the coronary stent industry. Prior to volume-based procurement, the market size for coronary stents was only around RMB 16 billion. Unlike innovative drugs in China, where a single drug can easily generate sales exceeding RMB 10 billion, for medical devices, a market at the RMB 10-billion level is already considered a large market.

Economic Observer Health: What is the core KPI for this year?

Zheng Lingxiao:Divided into two parts, with a primary focus on regions where the product has already been included in the medical service pricing catalog. In these regions, KPI targets will strictly prioritize driving hospital listing. The Marketing Department is tasked with strengthening the brand association between RaysightMed and FFR-CT, with the subsequent objective being to establish top-of-mind awareness: whenever FFR-CT is mentioned, RaysightMed must immediately come to mind.

In regions not yet included in the official pricing catalog, we continue to conduct expert advocacy, partner with third-party imaging centers, and provide services to out-of-hospital patients.

Economic Observer Healthcare: How do you view future competition? With the pace of regulatory approvals accelerating, could new competitors emerge soon, and is there a possibility of a price war?

Zheng Lingxiao:This is actually something we constantly contemplate and discuss. I still believe that for the development of the entire field, increasingly substantial investment is required; if only a single company contributes, the impact will remain highly limited. The medical industry is fundamentally different. A product cannot simply be launched immediately upon completion—it must undergo clinical validation and navigate the regulatory review process.

Basically, based on our current assessment of companies operating in this field, securing regulatory approval will likely require another two to three years. This indicates that we actually have a strategic window of approximately two years. Capitalizing on this window, we must prioritize market development, market access, and sales.

Furthermore, from the perspective of customer relationships, securing regulatory approval will not automatically guarantee market access in the future. It is actually highly difficult for competitors to penetrate our well-established network of clinical experts and partner hospitals. Therefore, we do not intend to engage in price wars going forward.

Economic Observer Healthcare: In terms of milestones, is our current progress aligned with our expectations, and what does the envisioned development roadmap look like?

Zheng Lingxiao:It is largely consistent. Prior to founding the company, each of the three founders had accumulated approximately a decade of technical expertise in related fields and possessed a thorough understanding of the industry. Consequently, subsequent stages such as product packaging and integration progressed very rapidly.

Our strategy is equally clear: first, bring one product to market to validate our capabilities, complete the entire regulatory approval process, and then advance to commercialization. For the first two years, we focused exclusively on the FFR-CT product, with the entire team fully dedicated to it. During this period, numerous business inquiries came our way, with hospitals frequently approaching us to ask whether we would consider developing liver cancer-related products or similar solutions.

Currently, our focus is twofold: advancing the commercialization of FFR-CT on one hand, and continuing to drive product market launches on the other. Over the next two to three years, we anticipate launching 5 to 6 products in the cardiovascular and cerebrovascular fields.