MedGPT: China's First Medical Large Language Model Redefines Digital Healthcare

In early 2023, a chatbot suddenly became a “global star”: ChatGPT.

This tool, developed by the U.S.-based artificial intelligence research company OpenAI, has achieved another leap forward in the history of machine-based question answering. Compared with existing search engines or voice assistants such as “Xiao Ai,” ChatGPT does more than simply retrieve information and provide basic responses; it understands human language and context, solves problems, and generates creative content. Numerous users have tested the system and found that ChatGPT handles tasks ranging from novel writing to programming with ease.

Simply put, it can think. As a result, upon its debut, ChatGPT first sparked a global wave of registrations and usage, drawing exclamations of “amazing,” and then its intelligence triggered widespread anxiety about job displacement across humanity.

Behind this ambivalence lies an elusive truth for humanity: maintaining a high level of professional excellence throughout the course of action.

Fortunately, the current version of ChatGPT has not yet had a significant impact on human jobs. However, its emergence can indeed play an important role in certain fields, such as providing professional assistance to users in legal and financial domains, and offering real-time support to students in the educational sector.

The most critical aspect lies in its application within the healthcare sector. First, such applications directly impact patients’ lives and health; second, the related products significantly influence the development of physicians’ professional competencies.

On April 28, WeDoctor Group unveiled its self-developed AI product, MedGPT, marking the first large language model-based AI product announced in China and the inaugural AI solution designed specifically for serious medical care.

The emergence of MedGPT is highly likely to comprehensively influence the medical awareness level of the Chinese public.

In September 1990, three students at McGill University in Montreal, Canada, wrote a program on bulky computers. This program would periodically connect to FTP hosts on the internet and download complete directory listings of all public files.

When users provide search terms, files matching those terms are retrieved and presented to them. This is widely recognized as the earliest form of search engine within the industry, built upon the foundation of the World Wide Web and information abundance. Its user base primarily consisted of individuals with basic computer literacy, such as students and university faculty. The use of search engines was closely associated with higher educational backgrounds.

Subsequently, as the Internet became increasingly rich and mobile networks gradually matured, user queries to search engines evolved from simple keywords to complete sentences. In China, many people instinctively think to “Baidu it” when encountering difficulties, a testament to how search engines have permeated every corner of daily life.

Search engines are continuously lowering the barriers for users. Today, people can perform searches via photo recognition or voice commands, eliminating the need for text-based input altogether.

Since search engines can already address people’s questions with such ease, why has the emergence of ChatGPT still sparked excitement across all of humanity?

The most significant difference is that ChatGPT, as an artificial intelligence product, proactively learns and updates, possessing the ability to understand natural language and contextual information.

For example, if you ask a search engine to “write a story about a witch flying in the sky,” it will retrieve all content containing the keywords “witch” and “flying in the sky,” whereas ChatGPT will generate a relevant story.

For search engines, ChatGPT has achieved a qualitative leap; it not only possesses the ability to filter content but also to recreate it. It focuses on problem-solving.

Thus, GPT-like products are a god-tier tool for industries across the board.

For a vast market like China, there is a particular need for genuine AI products to deliver convenient and professional services to regions with slower economic development. This requires such products to understand the connotations of Chinese culture, patterns of expression, and the developmental dynamics and trends of local industries. The ability to fully comprehend the Chinese language alone disqualifies many Q&A-based products.

Taken together, these requirements all suggest that developing AI-based products is no easy task.

Difficult as it may be, curiosity about new technologies has always driven people to explore uncharted territories. Currently, some players are taking on the challenge of the medical field, which demands higher professional expertise and involves more obscure knowledge.

On April 28, WeDoctor Group unveiled its self-developed AI product, MedGPT. According to publicly available information, the product is already capable of supporting initial consultations for nearly 3,000 diseases. It has also achieved breakthrough capabilities, such as multimodal dermatological diagnosis based on patient-uploaded images of skin abnormalities in the field of dermatology. Furthermore, it supports continuous, free-flowing dialogue between doctors and patients, marking a significant breakthrough in addressing a major industry challenge.

Given the corporate DNA and background of the company behind the product, the emergence of MedGPT appears to be a natural progression.

MedGPT is from Medlinker. As the name suggests, this is a doctor-centric internet hospital founded in 2014.

To date, we have accumulated nearly a decade of experience in the internet healthcare sector, which is no small feat. Those familiar with the developments in China’s healthcare industry understand that around 2014 marked the inaugural year of internet healthcare, when a myriad of digital health products emerged. However, as time has passed, many of these companies have faded into obscurity.

Yet Medlinker has remained a constant presence. According to official disclosures, the Medlinker platform already boasts over 1.5 million registered physicians and serves 20 million patients, indicating that Medlinker has penetrated to the core essence of doctor–patient connectivity in China. This represents the most valuable aspect of GPT-like products. It is on this foundation that the product can deliver effective responses within the correct semantic context.

Wang Lei, head of Medlinker’s Digital R&D Center, once stated, “We may not be the strongest AI technology company at present, but we are certainly the one that best understands how to leverage AI technology to better serve physicians and patients.”

"Staying committed to the healthcare sector may be the foundation of Medlinker's confidence in developing its medical GPT products."

Over the past decade, as a healthcare service provider with an internet-native foundation, Medlinker has placed significant emphasis on building and expanding the digitalization of doctor-patient interaction information on its platform. This has enabled the company to establish a complete closed loop encompassing prevention, diagnosis, treatment, and rehabilitation, thereby forging a robust doctor-patient relationship chain and disease management capabilities that are rare within the industry.

It is worth noting that, as a product that gained prominence alongside the rise of internet popularity, Medlinker has remained steadfast in its commitment to serious medical care. In 2021, Medlinker successively released standard operating procedures (SOPs) for multidisciplinary online disease management and became the first company in the industry to establish an academic committee led by academicians; in 2022, Medlinker also participated in the formulation of the "Expert Consensus on HIV-Related Medical Services Provided by Internet Hospitals."

These achievements stem from Medlinker’s continuous exploration of whole-process disease management models. It is understood that Medlinker has currently expanded into common therapeutic areas such as oncology, cardiovascular and cerebrovascular diseases, and diabetes, developing nearly 140 standardized disease management protocols and holding close to 100 patents or software copyrights.

Over the past decade, Yilian has deeply rooted itself in the healthcare services sector, uncovering key industry dynamics and developing a thorough understanding of the healthcare landscape, physicians, and patients. Leveraging its strong foundation in internet technology, Yilian was able to rapidly integrate resources to develop MedGPT—a solution primarily designed for primary care physicians and general patients—when there was a growing public demand for higher-quality medical interaction tools.

Its greatest significance lies in delivering the latest and most authoritative insights on a specific disease area to users from a medical perspective, thereby helping them address health-related concerns.

Having discussed the entry barriers for medical GPT products, we must return to the core question of demand: How important is a high-quality medical GPT product to us?

The other side of this issue is how much we pursue health.

Standing at the starting point of the post-pandemic era, “uncertainty” has become a source of anxiety for nearly every industry and all segments of the population. Amid this anxiety, people’s sense of self-protection has continued to strengthen. In January this year, the People’s Bank of China reported that RMB deposits increased by 6.87 trillion yuan in that month, an year-on-year increase of 790 billion yuan, hitting a record high for the same period historically. Savings represent people’s effort to enhance the “certainty” of their lives.

This holds true in the health sector as well, especially since the pandemic has made so many people acutely aware of the fragility of life. The difference lies in the fact that alleviating people’s anxiety over “uncertainty” in healthcare requires greater societal efforts, such as ensuring more abundant and high-quality medical resources, establishing a more robust emergency care system, and providing more professional and comprehensive health information.

This cannot be achieved overnight, and in regions with weaker economic development, people have less control over their healthcare.

In this context, the role of GPT-like products will become increasingly prominent. For instance, Medlinker has stated in its future application outlook for MedGPT that it aims to leverage AI capabilities to serve primary care physicians, helping doctors in remote areas improve disease diagnosis and treatment. According to Medlinker’s current test results, MedGPT’s diagnostic outcomes show a concordance rate of over 97.5% with existing diagnoses.

From the patient’s perspective, having access to a professional medical GPT product inevitably provides an additional layer of health security. As individuals increasingly seek greater control over their health, the value of such products continues to grow.

On another front, a professional medical GPT product will also play a positive role in China’s existing healthcare system. In a country as populous as China, with 1.4 billion people, healthcare reform has long been recognized as a particularly challenging endeavor. From the national reimbursement drug list and centralized drug procurement to hospital cost containment, efforts have predominantly focused on supply-side reforms, reflecting a singular directional approach.

AI technology can play a significant role in rapidly promoting the decentralization of high-quality medical resources and controlling health insurance costs, particularly if AI doctors pass double-blind tests. In other words, high-quality medical GPT products may have a positive impact on the structure of society as a whole.

MedGPT is currently undergoing intensive double-blind testing, with preliminary results expected in May.