Medical payment reform is moving into uncharted territory, and Maxin Health is exploring an AI "barrier-breaking experiment".
Recently, the World Artificial Intelligence Conference (WAIC) was held in Shanghai, attracting top figures in the AI circle. An increasingly clear trend has emerged: the main battlefield of AI is shifting from the competition of large - model parameters to in - depth engagement with industries, seeking value realization and solving real - world problems.
"AI + Healthcare" is one of the core sectors of this year's WAIC. Nowadays, the focus of industry discussions is no longer the performance scores of models, but how AI can effectively empower real - world scenarios such as diagnosis and treatment, payment, and new drug R & D, and complete the commercialization cycle.
There has long been a structural dilemma in the medical system known as the "medical impossible triangle": it is difficult to achieve high - quality services, wide - scale population coverage, and low costs simultaneously. However, with the accelerated maturity of AI and big - data capabilities and the continuous enhancement of the industry platform's understanding of scenarios, it is expected to break the shackles of the triangle, promote medical payment reform, dynamically optimize the reconstruction of medical resources, and provide the public with higher - quality inclusive medical services in the future.
Notably, in this industrial transformation, industrial platform companies with both technology integration capabilities and scenario understanding are emerging. Taking the medical payment segment as an example, Meixin Health, a Shanghai - headquartered technology - enabled diversified payment platform company in the pharmaceutical and medical field, is trying to use AI as the "underlying operating system" for reconstructing the medical payment ecosystem. Starting from hardcore scenarios such as claims settlement, it deploys AI Agent clusters to gradually reconstruct the value chain among "medicine - healthcare - insurance".
"This year's WAIC has pushed 'AI + Healthcare' to the center stage, indicating that large - model technology has reached the stage of industrial implementation. For enterprises, it is crucial whether AI can boost efficiency, optimize costs, and achieve win - win results in the ecosystem. Only when technology is deeply rooted in scenarios and serves people can the vision of 'global cooperation in the intelligent era' be truly realized." Qu Yuqi, the vice - president and head of the strategic innovation department of Meixin Health, said at the WAIC site.
Currently, against the backdrop of the accelerated integration of basic medical insurance and commercial insurance, the industry is closely watching how to use AI to solve the structural contradictions in the medical payment system. How will Meixin Health's "AI + Medical Payment" intelligent new infrastructure complete the closed - loop cycle from data integration, model iteration to scenario verification?
01. When AI Moves from "Showcasing Skills" to "Infrastructure Building"
Digitalization and intelligentization are among the trends in the technological evolution of the medical field. Over the past decade, the exploration of "AI + Healthcare" has been evolving, moving from single - point breakthroughs to systematic integration. During the stages of AI medical imaging and auxiliary diagnosis, enterprises actively improved the scores of their algorithms in international authoritative competitions, competing with human doctors in terms of speed and accuracy in single - point tasks such as identifying lung nodules and coronary heart disease.
Soon, the industry realized that algorithmic capabilities alone were insufficient to form a commercial closed - loop. The competition then entered the stage of vying for data and compliance. The focus of enterprises shifted from "performance scores" to who could cooperate with more top - tier hospitals to obtain high - quality, precisely labeled data and who could be the first to obtain the medical device registration certificate issued by the National Medical Products Administration. This certificate is an essential "qualification" for AI products to enter hospitals and achieve commercialization.
However, in the era of strict medical insurance cost control and in - depth medical payment reform, AI tools at the auxiliary diagnosis stage, even with the entry permit, still struggle to integrate into the diagnosis and treatment process and are unable to systematically solve the efficiency and cost problems of the medical system. Therefore, with the rise of large - model technology, AI medical practice has entered the second half of the competition, which focuses on "scenario and ecosystem" capabilities. The focus of competition is no longer the quality of a single tool, but who can integrate AI throughout the entire patient treatment process, from pre - diagnosis, in - diagnosis to post - diagnosis, and finally to payment.
Payment is the key to verifying the business model of medical AI and forming a value closed - loop. Therefore, it is an inevitable choice to transform AI capabilities into stable, fair, and scalable "infrastructure" to empower this core scenario.
02. The AI Revolution Enters the Medical Payment Field
For years, patients, pharmaceutical companies, and insurance companies in the medical payment chain have each faced difficult challenges. For patients, high drug costs and cumbersome claims processes have led many families to "fall into poverty due to illness or return to poverty due to illness". Pharmaceutical companies face difficulties in new drug access and commercialization after a long R & D cycle. Insurance companies struggle with high marketing and claims settlement costs and insufficient risk - control refinement.
The pharmaceutical and insurance industries have high cognitive barriers and face the problem of information asymmetry, compounded by data silos. As a result, it is often difficult for insurance companies and individual pharmaceutical companies to interact efficiently. Therefore, the market needs a diversified payment platform that can connect all parties, with both the ability to manage the pharmaceutical supply chain and the ability to design diversified payment solutions for patients and support innovation in insurance product sales to bridge the gaps in the "medicine - healthcare - insurance" industry.
In response to this industry demand, a group of platform - type enterprises with cross - domain integration capabilities have begun to emerge. They attempt to use technology as a link and the ecosystem as a support to fill the gap in the collaboration among "medicine - healthcare - insurance". Meixin Health is one of the explorers and practitioners in this field.
Since its establishment in 2017, Meixin Health has been committed to this direction, and now its scale effect is beginning to show. According to its prospectus released on June 30, as of the end of last year, the platform had served over 1.6 million patients, with the total relevant medical payment volume (GPV) reaching 39.7 billion yuan; it had cooperated with over 90 insurance companies, supporting approximately 393 million insurance policies; and it had partnered with over 140 pharmaceutical companies, including 90% of the world's top 20 pharmaceutical companies.
While its revenue scale has expanded, Meixin Health's marketing expenses have significantly decreased, which is inseparable from the construction of its AI intelligent central platform. Within Meixin, this AI central platform named "mind42.ai" is defined as the underlying operating system for coordinating the "medicine - healthcare - insurance" ecosystem.
"Since 2024, Meixin Health has been trying to comprehensively integrate the value chain of the pharmaceutical and healthcare industry, covering the entire process from product design, risk control, sales to claims settlement on the insurance side, as well as health management and data - value application on the patient - service side," Qu Yuqi introduced. "Currently, we have found application scenarios for intelligent agents in key nodes such as claims settlement and customer service. Although it is still in the early stage, in the future, with the improvement of more intelligent agents, we hope to use the intelligent central platform mind42.ai for unified task scheduling to connect all nodes in the long - value chain."
The decision - making chain in medical payment is complex and requires the integration of multi - domain knowledge systems. According to the prospectus, the AI intelligent central platform, based on an independently developed base of advanced open - source models, has accumulated 385 million detailed claims data assets; it integrates vertical - domain knowledge such as clinical medical pathways, insurance terms, and drug indications to build a medical payment knowledge graph. Being "knowledgeable in both medicine and insurance" makes it possible for AI to automate core processes and improve decision - making efficiency.
In the process of processing and scheduling data, services such as claims settlement, drug purchase, and health management provided by the platform will continue to generate structured data, which are fed back to the AI model in real - time for training and optimization, forming a "data flywheel" that makes the mind42.ai model more intelligent and gradually builds a barrier that is difficult to replicate in the short term.
Currently, data is becoming a "flowing resource" with more dynamic value. In the medical payment scenario, insurance companies usually handle claims after the event, but now we are trying to use AI and big data to shift it towards "pre - event prevention and in - process intervention". For example, for patients with chronic diseases, we can identify risks in advance and proactively conduct health - management interventions. In the future, we may even achieve dynamic insurance pricing," Qu Yuqi shared.
03. AI Agents "Fight in Groups"
Medical payment is essentially a financial issue, and the tolerance for decision - making errors in the scenario is extremely low: over - payment will increase the costs of insurance companies and the platform; under - payment will damage the interests of patients. This particularity determines that it has much higher requirements for decision - making accuracy than ordinary fields.
Previously, the claims settlement of health insurance highly relied on multi - domain professional collaboration: it required staff with professional medical backgrounds to carefully review patients' medical records, diagnosis and treatment reports, genetic testing reports, etc., to judge the rationality of the treatment path, especially for innovative drugs with new and narrow indications, which required strong medical professional judgment; at the same time, it also needed insurance experts to interpret and match complex insurance terms and risk - control and compliance experts to manage risks. Although this multi - role collaborative model can ensure accuracy, it also brings industry pain points such as cumbersome processes and low efficiency.
This places strict requirements on the decision - making ability of AI. It needs to "understand the diagnosis and treatment logic like a medical expert, calculate sharing rules like an insurance actuary, and identify fraud like a detective". Obviously, traditional manual methods or single general - purpose AI are difficult to handle such complex tasks, which highlights the value of AI Agent clusters: through the "group fight" of multiple intelligent agents, they can efficiently and collaboratively complete comprehensive tasks that a single entity cannot handle.
Following this idea, Meixin Health's current AI Agent cluster is building a collaborative network covering "front - end interaction, middle - platform decision - making, and back - end fulfillment". mind42.ai serves as the "commander - in - chief" of the intelligent agent matrix, automatically calling the corresponding intelligent agents according to scenario requirements to achieve efficient collaboration.
At the front - end, C - end users interact through the "Xiaofu Intelligent Assistant" to handle customer - service content such as insurance - term inquiries and drug searches; at the claims - settlement and payment layer, centered around the "ClaimMaster", users can upload materials with one click, and the intelligent agent can automatically complete document classification, key - information extraction, and calculate the claim amount.
Meixin Health's Xiaofu Intelligent Assistant
Meixin Health's choice to use "claims settlement" as the first - landing scenario for AI Agents is not accidental. This scenario is not only a core pain point in user experience due to its cumbersome and time - consuming process but also highly compatible with the technical capabilities of intelligent agents because of its high data integrity and strong repeatability of operation steps.
"When a patient submits claims materials, the traditional method requires claims - settlement staff to carefully review each page. The most complex claims materials can have thousands of pages, and an ordinary case usually has dozens of pages," Qu Yuqi explained. "However, after applying the claims - settlement intelligent agent, with a series of technologies such as OCR recognition and multi - modal compensation, the overall claims - settlement time can be shortened from the original one - day agreement to within 10 minutes at the fastest."
It is reported that currently on the Meixin platform, the AI - empowered medical review covers 60%, significantly solving problems such as inefficient material processing, rough disease - course analysis, and delayed term interpretation in the health - insurance drug claims - settlement process. While improving the patient experience and reducing the financial burden on families, it also saves Meixin's operating costs.
In the future, as the task - handling ability of the AI Agent cluster becomes more mature, the interaction scenario between Meixin users and multiple AI intelligent agents may be as follows: when a user enters the platform and asks about suitable health - insurance products, Xiaofu will recommend corresponding insurance or diversified payment products according to his physical condition (healthy or with a disease). After the user purchases insurance and needs to make a claim, the system will call ClaimMaster to handle his drug - claim needs. At the same time, the back - end drug - delivery management system (including the intelligent robots in the commercial - insurance intelligent cloud pharmacy) can complete the unpacking, sorting, and delivery of drugs.
Based on its AI middle - platform capabilities and exploration in intelligent agents, Meixin Health is trying to extend its capabilities to more scenarios, such as providing health - management services for chronic - disease patients and helping pharmaceutical companies with patient management. Deep integration with customers and providing more value - added services are not only a natural extension of existing capabilities but also lay the foundation for exploring more value - growth points.
04. The Boundaries and Future of Medical AI
Recently, the Research Institute of China Merchants Bank pointed out that AI - enabled medical payment is evolving from "process automation" to "intelligent risk control" and "actuarial - driven development". This is not only an improvement in efficiency but also a reshaping of the "risk - pricing management" logic in the health - insurance industry.
This judgment coincides with Meixin Health's industry observation, and it has more acutely captured the deep - seated transformation of the role of the technology platform. Qu Yuqi said, "Medical payment is upgrading from human - dominated to intelligence - driven. The payment platform is expected to become a 'participant in the system' and participate in formulating new industry standards, such as payment processes, claims - settlement norms, and even drug access methods. In the future, we hope to use technology to promote the dynamic optimization of medical resource allocation from static misallocation."
Indeed, more personalized and precise AI - enabled medical payment and health management highly depend on the integration of "full - link data". Currently, patients