Navigating the Strictest-Ever Regulation: How Internet Healthcare Platforms Achieve Compliance

Recently, the Medical Administration and Hospital Management Bureau of the National Health Commission released the “Detailed Rules for the Supervision of Internet-Based Diagnosis and Treatment (Draft for Comments)” (hereinafter referred to as the “Rules”). Key provisions—such as the appropriate use of artificial intelligence (AI), the clarification of documentation and definitions for follow-up consultations, and the prohibition of linking physicians’ income to pharmaceutical sales—have become focal points of industry attention.

Since 2020, policies issued by the National Health Commission regarding internet healthcare have primarily been supportive and encouraging. However, the current Detailed Rules emphasize regulatory constraints. Through discussions with multiple experts, VCBeat has learned that the Detailed Rules will have limited impact on internet hospitals led by public hospitals, but rather reflect further standardization of internet healthcare enterprises. This is because public hospital-led internet hospitals, building upon existing policies and incorporating elements from the regulatory framework for physical hospitals, are already subject to relatively strict management requirements. In contrast, internet healthcare enterprises lack comparable experience in medical management, necessitating the establishment of more detailed regulations.

Against this backdrop, how should internet healthcare companies operate in compliance with regulations? This issue will be explored from the following three aspects.

Article 13 of the Detailed Rules stipulates that physicians must undergo real-name authentication prior to patient consultations to ensure that they personally conduct the consultations. Other personnel, artificial intelligence software, and similar entities are prohibited from impersonating or substituting for physicians in conducting consultations.

Previously, the Yinchuan Internet Diagnosis and Treatment Service Specifications (Trial) and the Sichuan Province Internet Diagnosis and Treatment Management Regulations (Trial) had already established corresponding requirements: artificial intelligence and other technologies shall not be used to completely replace physicians in conducting consultations, writing medical records, prescribing medications, and other diagnostic and treatment activities. The Hainan Province Internet Diagnosis and Treatment Quality Management Guidelines (First Edition) also explicitly prohibit the use of artificial intelligence-assisted programs (AI) to replace physicians in issuing prescriptions.

The wording in several documents varies slightly, but their core message consistently emphasizes that AI should be used only as an aid and not as a substitute.

Currently, AI has been applied in multiple scenarios of internet healthcare. The most commonly used AI products mainly fall into several categories:

First, AI-powered triage or patient guidance systems enhance the patient experience. After patients input information such as affected body areas and symptoms based on system prompts, the system recommends appropriate medical departments or directly matches them with physicians in those specialties. According to a survey of over 7,000 healthcare users conducted by VCBeat Research Institute in its “2021 White Paper on Smart Hospital Innovation,” 32.3% of respondents were unsure which department to visit before seeking care, and 34.1% did not know which doctor to choose. Similar challenges exist in online healthcare settings; therefore, AI-based triage and guidance can significantly improve the efficiency of matching patients with appropriate healthcare providers.

Second, AI-powered pre-consultation systems, which interact with patients through standardized questions to collect clinical information, thereby preventing physicians from spending excessive time on simple, repetitive inquiries during consultations.

Third, AI-assisted decision-making systems, which are similar to the Clinical Decision Support Systems (CDSS) used by physical medical institutions, provide assistance to physicians in making diagnostic and treatment decisions.

Furthermore, AI can also be applied in post-diagnosis follow-up, medical quality control, and other areas.

The appropriate application of AI can enhance the overall efficiency of internet-based diagnosis and treatment services. However, phenomena such as AI completely replacing physicians in patient consultations or pharmacists in prescription reviews could lead to consequences ranging from financial losses for patients and digital health platforms to severe setbacks for the industry as a whole. How should this double-edged sword of AI be wielded? For digital healthcare enterprises, it is imperative to strictly maintain the boundary between “assistance” and “replacement.”

Currently, many companies are applying AI to scenarios that assist physicians while maintaining strict control over its usage boundaries. For instance, Chunyu Doctor was among the early adopters to integrate artificial intelligence into the patient–physician matching process. Zuoshou Doctor’s intelligent triage and smart pre-consultation systems can enhance patients’ consultation efficiency and help physicians gain prior insight into patients’ conditions, thereby assisting in clinical analysis and risk assessment.

Ping An Health has developed an AI-assisted diagnosis and treatment system, featuring diagnostic models covering over 2,000 diseases, to help physicians across 22 departments improve the efficiency and accuracy of daily consultations. However, these improvements in efficiency and quality are built on two fundamental premises: First, all physicians at Ping An Health who conduct disease diagnoses and prescribe medications hold the corresponding professional qualifications and have completed their practice registration or multi-site filing with the internet hospital. Second, the system solely provides information collection and decision support for physicians based on databases, deep learning algorithms, and feedback loop learning. Ultimately, whether issuing diagnostic and treatment recommendations or prescribing medications, all actions are carried out by the physicians themselves based on the patient’s actual condition.

Furthermore, internet healthcare platforms can exercise control over procedural details. For instance, upon transition from the AI triage and guidance phase to direct consultation by a physician, visual or textual prompts may be displayed in the chat interface, while maintaining comprehensive records of all communications throughout the entire process. This approach not only enhances patients’ perception of the authenticity of the consultation but also complies with the requirements stipulated in the Detailed Rules for full-process documentation and traceability of diagnosis and treatment activities.

Proof of follow-up visits and their assessment are also among the key points that have drawn widespread attention in these Detailed Rules.

Article 18 of the Detailed Rules stipulates that patients shall provide medical records with a clear diagnosis, such as outpatient medical records, inpatient medical records, discharge summaries, and diagnostic certificates, when seeking medical attention. The attending physician shall determine whether the patient meets the conditions for a follow-up visit and collect paper or electronic documentation proving that the patient has been definitively diagnosed.

Industry experts believe that given the highly specialized nature of the healthcare sector, physicians should flexibly determine the criteria for follow-up consultations in practice, rather than adhering to rigid standards. The new regulations fully respect physicians’ decision-making authority. However, internet healthcare platforms should not only conduct online follow-up consultations in accordance with this principle but also improve their information systems to collect and retain patient data as required.

Similarly, the determination of whether an in-person visit is required is left to the physician’s judgment. The Detailed Rules stipulate that if a patient’s condition changes, the physician determines the current consultation to be an initial visit, or other circumstances arise that make internet-based diagnosis and treatment inappropriate, the attending physician shall immediately terminate the online consultation and direct the patient to seek care at a physical medical institution.

In fact, when equipped with appropriate and accurate information, it is not difficult for physicians to use their professional judgment to determine whether a consultation constitutes an initial visit or a follow-up. For internet healthcare platforms, the greater challenge lies in establishing a service workflow that seamlessly integrates online and offline care, ensuring continuity between initial consultations, follow-up visits, and other necessary patient interventions.

Over the past few years, the internet healthcare industry has undergone a rapid and extensive build-out of online service systems; however, offline service capabilities remain relatively weak. Taking laboratory and diagnostic testing as an example, these services must be performed in person, and their results serve as critical evidence for disease diagnosis, yet internet healthcare platforms have had limited involvement in this area. In the last two years, major internet healthcare platforms have accelerated the integration and development of offline service capabilities. Key approaches include establishing internet hospitals in multiple regions, partnering with public hospitals to operate internet hospitals, collaborating with third-party medical testing institutions, building or acquiring physical hospitals, and intensifying brand promotion through offline channels.

For example, Ping An Health has built an O2O service ecosystem that integrates online and offline processes end-to-end. A defining feature of this model is its use of Ping An Health’s family doctors as guides to bridge online and offline services. By leveraging medical institutions, health management organizations, consumer healthcare providers, and diagnostic testing facilities, it delivers four types of services to users: online consultations, at-home offline care, in-person visits to offline facilities, and offline delivery. Additionally, it establishes personal health records for users to ensure continuity of care.

Strengthening the integration and strategic deployment of offline medical resources enables internet healthcare platforms to obtain more comprehensive and accurate patient consultation data, thereby reducing medical risks and enhancing service quality.

The Detailed Rules further stipulate that medical institutions conducting internet-based diagnosis and treatment activities shall strictly comply with prescription management regulations, including the Measures for the Administration of Prescriptions, strengthen medication management, and prohibit practices such as aggregated prescription data collection for commercial purposes and retrospective prescription issuance. The personal income of healthcare professionals shall not be linked to revenue from pharmaceuticals or medical tests.

In fact, the decoupling of drug sales from physicians’ income is a regulatory policy that has been implemented offline for many years. However, as similar trends have emerged in the online sector, the Detailed Rules specifically reiterate this requirement.

In managing the relationship between medical services and pharmaceuticals, approaches can be adopted from either an operational or a performance evaluation perspective. For instance, Haodf.com does not engage in drug sales operations; instead, it partners with multiple online pharmacy platforms. After physicians issue prescriptions, patients are free to choose which platform to purchase their medications from, thereby eliminating conflict-of-interest concerns associated with prescription-driven revenue streams.

Of course, even offline hospitals have in-house pharmacies. Internet healthcare platforms are not prohibited from engaging in pharmaceutical retail operations provided they possess all necessary qualifications; rather, they employ mechanism design to prevent the occurrence of drug kickbacks. It is understood that Ping An Health has also taken steps in terms of operational strategy and physician incentive mechanisms to separate medical services from drug sales.

Currently, Ping An Health has established synergies with insurance companies and financial institutions, building a bridge between suppliers and payers within the medical industry chain. It provides healthcare service products and full-lifecycle management to insurance or financial clients and employees of large enterprises, creating a multi-party beneficial HMO (Health Maintenance Organization) model. Specifically, Ping An Health sells membership-based products to insurance companies and financial institutions, integrating online and offline resources to deliver targeted and appropriate services to users. This approach aims to achieve the dual goals of improving health outcomes and controlling costs, thereby avoiding over-treatment.

Regarding the incentive mechanism, physicians on the Ping An Health platform receive compensation comprising a base salary and performance-based pay. The performance-based component is assessed primarily based on metrics such as patient volume, patient satisfaction ratings, and quality of medical services, with no linkage to pharmaceutical sales.

Overall, the Detailed Rules impose higher requirements on the internet healthcare industry. Although the document reflects stringent regulation, it is markedly different from the 2017 Administrative Measures for Internet Diagnosis and Treatment (Trial) (Draft for Comments) and the Opinions on Promoting the Development of Internet Medical Services (Draft for Comments).

The regulatory approach reflected in the two 2017 documents underscored industry uncertainty, whereas the current Detailed Rules were formulated on the basis of strong national affirmation and encouragement of internet healthcare. Their purpose is to standardize industry practices and promote healthy, orderly development. Industry insiders also widely believe that, following the implementation of the Detailed Rules, the sector is poised to undergo a process of survival of the fittest, entering a more virtuous cycle.