AI Healthcare Going Global: The Commercialization Dilemma of AI in Medicine
-
In 2015, at his entrepreneurial mentor's home, Chen Kuan wrote down four words: 'cross-border, global, integration, innovation.' These words encapsulated the founding vision of Infervision Medical and aligned with the future of the AI healthcare industry.
This year, internet healthcare reached a life-or-death edge, with a batch of O2O-model internet startups racing against time for survival. Their investors discovered that services were merely online appointment bookings, and the so-called 'full life-cycle' offerings had no buyers at all.
Across the ocean, the internationally renowned medical imaging company Enlitic was established and developed software capable of identifying malignant tumors from X-ray and CT scan images. In 2015, Mount Sinai Hospital in the United States utilized an AI technology called Deep Patient, which performed exceptionally well in analyzing the medical record data of 700,000 patients at the hospital.
This phenomenon was noticed by Chinese investors. Soon after, the most perceptive angel investors observed, 'Hmm? AI in healthcare seems to be emerging.'
The first to take action was Lü Chuanwei, who is also widely known as the founder of Kuaidi.
In addition to individual investors placing their bets, a wave of early-stage institutions swiftly entered the arena. After securing angel-round funding, Infervision became a pioneer in medical imaging + AI and began advancing. Gradually, more companions joined the journey, revealing the opportunities in healthcare + AI to a broader audience. This not only attracted a surge of investors but also led to a rapid influx of market participants. On one hand, more players brought vibrancy to the industry, but on the other, increased competition intensified the race.
Who can go the distance is not the key question, the core proposition is: who is faster?
2017 was undoubtedly the most booming year for AI medical imaging in China. The year before, investors were still on the sidelines, and the year after, the capital winter tightened investors' purse strings. It was precisely in this middle year that AI medical imaging reached its peak.
Data shows that in 2017, the medical artificial intelligence industry developed rapidly. There were nearly 30 publicly announced financing events in the industry, with total financing exceeding 1.8 billion yuan. Among them, companies such as Infervision, Deepwise, and 12Sigma each secured two rounds of financing within a year, with each financing round exceeding 100 million yuan.
This year, international media such as Forbes have enthusiastically praised China's AI rising star Infervision, calling it 'China's trailblazer in the global AI field'.
At that time, Tuixiang Medical began promoting the offline deployment of its products. According to incomplete statistics, that year, Tuixiang Medical collaborated with over 50 top-tier hospitals. Doctors at Wuhan Tongji Hospital, after trying the collaboration, found that Tuixiang Medical could complete image reading in just 5 seconds, giving it high praise and calling it the 'CT AlphaGo.'
In the summer of that year, Chen Kuan, CEO of Infervision, was invited to Silicon Valley to deliver a speech titled 'From Model to Product.' Chen Kuan shared his vision—leveraging deep learning to alleviate the shortage of radiologists in China's healthcare system.
According to his speech at the time, during this phase, Inference Medical's product deployed in several top-tier hospitals achieved a nodule screening diagnosis rate of over 90%, with daily growth. Comparative studies found that the smaller the nodule size, the more pronounced the accuracy advantage of Inference Medical's product became compared to first-line physicians' diagnostic performance.
Notably, Chen Kuan proposed a globalization plan at this time. He stated, 'While continuously optimizing its models, Infervision is also collaborating with partners such as GE, NVIDIA, Cisco, and Intel to establish a strong presence in the Chinese market through multiple channels. Infervision's journey began in China, and in 2017, we set sail to explore global opportunities.'
An AI medical imaging company's ambition to go global is closely related to the founder's vision, foresight, and ambition. From Chen Kuan's personal perspective, he studied from undergraduate to Ph.D. at the University of Chicago but chose to return to China to start a business just before graduation. This international perspective may be deeply connected to these experiences.
However, to be frank, although there are plans for overseas expansion, whether it's Infervision, Deepwise, or Shukun Technology, their main battlefield at that time was still the domestic market. From the current perspective, before venturing into overseas markets, they still share a common issue: commercialization.
In the past few years, various competitors have successively secured funding. Apart from Infervision, companies like Shukun Technology, Eagle Eye Technology, Deepwise Medical, and Shanghai United Imaging have all launched their flagship products and captured their respective shares in the market.
But resources are limited while demand is high. Additionally, the 2018 "New Asset Management Regulations" have made investors more cautious. This has directly led to two phenomena: a more competitive landscape and a halt in capital support.
This means AI medical imaging projects must now rely on themselves to survive.
There was once an industry joke: "AI medical projects are so poor they only have technology left." As capital momentum wanes, investors have become more discerning, redirecting their attention to the innovative drug sector. For AI medical imaging companies, possessing algorithmic technology alone is insufficient—the key lies in addressing real-world clinical needs.
A critical issue is the lack of data. The digital transformation of China's healthcare system has only begun in recent years. Just a few years ago, patients had to purchase paper medical record booklets for one yuan during hospital visits, resulting in low usability of these records. Although this situation has improved in recent years, data remains siloed across different hospitals. This significantly increases the difficulty of accumulating high-quality datasets. Companies must establish partnerships with hospitals one by one and then use the collected data to train their AI products, highlighting the immense challenges and high human resource costs involved.
At the same time, AI-assisted medical imaging diagnosis faces stringent regulations, with AI medical imaging products classified as highly regulated Class III medical devices. These hurdles have significantly delayed the implementation of AI in medical imaging.
2020 marked a watershed moment for AI in medical imaging.
After a long wait, the relevant participants finally received their 'certificates'.
In 2020, China's National Medical Products Administration (NMPA) officially granted Class III certification to Infervision's AI-based pulmonary nodule detection system, marking the first pulmonary AI product to receive Class III approval from China's NMPA.
Early this year, Infervision Medical's products became the first in the world to receive FDA and PMDA certifications for AI in chest and lung CT imaging. At that time, Infervision also became the first and only AI medical company to hold approvals from the EU CE, Japan PMDA, US FDA, and China NMPA, gaining access to most major global healthcare markets. Later in 2023, Infervision Medical secured the UKCA certification, further solidifying its leading position with a 'Global Five-Certification' advantage.
The primary applications of AI in medical imaging include lesion identification and annotation, automatic target delineation for adaptive radiotherapy, and three-dimensional image reconstruction. After several years of development, medical imaging has now become the most mature application field of AI healthcare in China, with the most extensive applications in pulmonary, cerebral, ocular, fracture, and cardiovascular areas. Currently, there are over twenty approved Class III medical device certificates in these fields, with AI technology being most mature in pulmonary disease applications and having the highest number of approved products.
A fact is that for a long time, the core contradiction in the development of the AI medical imaging industry has mainly focused on commercialization progress. With 'certification,' companies can take a step closer to commercialization.
Last year, some professionals in the medical industry outlined the path for medical imaging AI: if following the tertiary hospital route, companies first need regulatory approval, then pricing approvals from provincial and municipal authorities, followed by out-of-pocket patient purchases. After 1-2 years of operation, the service would transition to insurance reimbursement. A rough calculation suggests that from obtaining regulatory approval to achieving profitability, companies would need a 4-5 year timeframe.
As mentioned by the experts above, obtaining certification is not sufficient—insurance payments are also necessary to achieve true profitability. However, is there really no alternative path?
Going global is the best path.
Beyond Infervision, a number of AI medical projects like Subtle Medical have already tasted success overseas.
According to market reports, Subtle Medical has secured nearly 100 million yuan in orders in just the first seven months of 2023. Currently, the company's global operations primarily follow a pay-per-case and recurring payment model with hospitals, successfully deploying its solutions in over a hundred leading tertiary hospitals in China, often through independent procurement projects. If this momentum is maintained while controlling costs, the company is expected to achieve breakeven this year and potentially turn profitable within two years, breaking the profitability challenges faced by medical AI companies.
In fact, Infervision may indeed be one of the earliest AI imaging companies to go global.
With an international imprint embedded in its startup DNA, Infervision Medical began building its overseas team in 2018, establishing subsidiaries in North America, Europe, and Asia-Pacific. The company continues to attract local talent to expand its regional coverage and has made continuous progress and achieved positive developments in areas such as product registration, overseas commercialization, and the formulation of international standards.
It is reported that Infervision Medical's products have obtained nearly 10 overseas market certifications, including EU CE certification, EU MDR CE certification, US FDA certification, Japan PMDA certification, and UK UKCA certification. The products are available in countries and regions such as China, Spain, France, Germany, Austria, Japan, Italy, Switzerland, Canada, and the United States.
Currently, Infervision's AI medical solution has been selected for the United Nations Global Drug Facility (GDF) procurement list, making it a global supplier for government sovereign fund purchases, aid fund procurement, and purchases by various international organizations. It is also among the first batch of AI products to be included in this global list. Through procurement by the United Nations Development Programme (UNDP) and the United Nations Office for Project Services (UNOPS), Infervision's AI medical solution has been successfully deployed in Belt and Road countries such as Uzbekistan, Moldova, and South Africa. The solution utilizes artificial intelligence to assist in disease screening, contributing to improving local tuberculosis screening and reducing mortality rates.
Additionally, Infervision's AI medical solutions have been successfully included in the UK NHS Shared Business Service (NHS SBS) framework and have consistently secured EU government centralized procurement. This demonstrates that Infervision's service and deployment capabilities have withstood the test of the international market.
Infervision is actively engaged in technological and product exchanges with countries along the Belt and Road initiative. The company has signed a memorandum of understanding with Ethiopia's Ministry of Science and Technology to jointly promote the implementation of AI in the African Union.
Infervision participated in the international exchange of the 'Medical AI Product Standards Development Group' organized by the WHO (World Health Organization) and ITU (International Telecommunication Union), actively engaging in the formulation of international standards.
By 2023, Infervision has achieved product coverage in over 1,000 medical institutions across 25 countries worldwide.
Why Go to Overseas Markets?
First, overseas markets are more receptive to paid models;
Similar to the enterprise services sector, after years of validation, the European and American markets recognize paid models. For example, medical imaging projects using SaaS payment models are accepted by hospitals and medical institutions, which have established payment habits. They prefer purchasing services to enhance efficiency and reduce high labor costs.
Second, the overseas medical insurance system is more comprehensive;
Compared to China's insurance industry, the number of people purchasing insurance abroad is significantly larger and more common. Taking the U.S. market as an example, without insurance products, people even hesitate to call an ambulance—due to unaffordable medical costs. However, it is precisely this well-established insurance system that enables them to access more comprehensive paid services.
Third, overseas markets have enormous demand.
For example, in the European market, with its developed economy and severe population aging, lung cancer ranks third among the most common cancers and is the leading cause of cancer-related deaths. Additionally, Europe is the second-largest market for CT scanners globally, with lung cancer accounting for nearly 20% of all cancer cases in European countries.
Europe has a population of approximately 500 million. If following the screening ratio in the United States, there would be about 14 million people eligible for early lung cancer screening, potentially creating a lung cancer AI market worth around one billion euros. Currently, Europe is placing increasing emphasis on early lung cancer detection, as identifying and treating lung cancer at an early stage could save substantial healthcare costs for the entire medical system.
In 2020, Infervision Medical obtained CE certification, illuminating its commercial expansion in Europe.
There are many such cases. For instance, during the fight against COVID-19, Infervision Medical's AI solutions for COVID-19 assisted medical institutions in multiple countries and regions, including Europe and Japan. The EU government also repeatedly procured Infervision Medical's AI solutions for COVID-19 in bulk.
In July 2021, when Infervision Medical completed its D2 round of financing, it was also included in the United Nations procurement list. This undoubtedly accelerated the advancement of Infervision Medical's globalization strategy. Behind this lies the implication that Infervision Medical can navigate the capital winter by walking on two legs—leveraging both the Chinese market and overseas markets, with multiple product pipelines deployed simultaneously.
Today, 'AI+Healthcare' has become a 'cure' for alleviating medical resource shortages in China and globally. Previously, due to unclear commercial pathways and unverified application scenarios, market understanding of AI healthcare was insufficient. However, as AI healthcare has progressed from initial emergence to rapid iteration and deep learning expansion, it has now officially entered the registration approval and market transition phase. Many forward-looking AI healthcare companies have developed deployable software solutions and realized tangible revenues, ushering in a phase of rapid development.
However, from a physician's perspective, an outstanding leader in medical AI should view the field through multiple lenses: technological development, business growth, healthcare needs, and economic returns. They understand that obtaining regulatory approval is not the ultimate goal—securing approval for a product with genuine market value is what truly matters.
As Chen Kuan himself said, "Medicine is an extremely rigorous profession, and doctors from different regions have varying habits when dealing with specific clinical issues. Only by staying close to doctors and clinical practice, obtaining the most authentic feedback from physicians, can we develop products that best fit the actual scenarios."
Stay true to your original aspiration, and you will achieve your goal.