"Zijing Zhikang" erhält fast 100 Millionen Yuan an Angel-Runden-Finanzierungen, um die Entwicklung und Implementierung des AI-Krankenhaussystems zu beschleunigen | Erstveröffentlichung von 36Kr
36Kr learned that "Zijing Zhikang" recently completed an angel round of financing of nearly 100 million yuan. The round was led by Xinglian Capital, with follow - on investments from Inno Angel and Shangshi Capital. The funds from this round will mainly be used for the research, iteration, and upgrading of the Zijing AI Hospital (Agent Hospital) system.
Zijing Zhikang was established in September 2024, incubated by the Institute for AI Industry Research (AIR) at Tsinghua University, initiated by Liu Yang, a professor in the Department of Computer Science at Tsinghua University and the executive dean of AIR, and seed - funded by Qingzhi Capital. It attempts to use cutting - edge large - model agent technology to develop a medical virtual world system and promote its application and optimization in the real world, thereby empowering smart healthcare.
The attempt to empower healthcare with AI has a long history. Many sub - sectors such as AI medical imaging and AI medical big data have produced unicorn companies. However, the iterative development of capabilities in this field has continuously faced pain points such as data asset compliance and product commercialization.
The core logic of the Zijing AI Hospital is to simulate the facilities and processes of a real hospital. In particular, it constructs highly anthropomorphic, widely distributed, and diverse AI patients to meet the initial training data requirements.
Furthermore, it develops AI doctors with self - evolution capabilities to provide users with convenient, affordable, and high - quality medical services, solving the "impossible triangle" in healthcare. Ultimately, it aims to integrate the entire cycle of users' pre - diagnosis, in - diagnosis, and post - diagnosis health management, constructing an "AI - driven smart healthcare closed - loop system" rather than being limited to plug - in AI medical tools.
To enable AI doctors to reach a high level, high - quality data and medical cases are first required for training. However, in the real medical world, there have long been challenges such as data silos, high difficulty in data acquisition, and complex data governance. Therefore, the core technology team of Zijing Zhikang has found an alternative way by using AI to synthesize some case data, attempting to solve the initial cold - start problem and constructing an "evolvable agent based on simulacra".
Specifically, with the help of "large models + medical knowledge base + a small number of case library samples", through multi - step inverse sampling, that is, determining the basic attributes such as age and gender corresponding to a disease based on medical knowledge, generating past medical histories, symptoms, and examination results, and then verifying through self - checking by the large model and annotation by human experts, cases are automatically synthesized. Combining the role - playing ability of the large model, the synthesized cases are transformed into AI virtual patients, allowing the evolvable AI doctors to conduct a complete diagnosis and treatment process with them and obtain feedback.
According to reports, the Zijing AI Hospital has currently constructed over 500,000 AI patients, covering different countries, age groups, and disease types, becoming an important supplementary path for training AI doctor agents. However, the team also said that the current synthetic data cannot cover all medical situations in the real world, especially complex cases of multiple comorbidities. Therefore, the team also values real - world data and feedback during the process of improving the model's capabilities.
The core of human doctors' ability to "become more capable with more consultations" lies in the accumulation of actual clinical experience. Enabling AI doctor agents to also have "self - evolution" capabilities is another core technology of Zijing Zhikang.
The team has designed a specific memory and reflection algorithm mechanism to allow AI doctors to accumulate "experience" in the closed - loop consultation process. The method of experience management is very important. It is necessary to verify its effectiveness and avoid over - generalization. After multiple rounds of verification, experiences of different orders will enter the database of the AI doctor agent and evolve into its "exclusive memory".
In the virtual world, AI doctors can conduct a large number of diagnosis and treatment practices in a short time, which means their evolution speed will far exceed that of human doctors. "Experiments show that the ability evolution curve of AI doctors has initially conformed to the Scaling Law: If AI patients meet the conditions of high anthropomorphism, wide distribution, and diversity, then the more AI patients an AI doctor diagnoses and treats, the stronger its ability will become."
According to Zijing Zhikang, the team has developed 42 AI doctors, with an accuracy rate of over 96% on the internationally authoritative MedQA dataset, exceeding the average level of human doctors.
In terms of product design, the current AI system developed by Zijing Zhikang has three ports: a patient - side APP, a doctor - side workstation, and a hospital system to achieve full - cycle closed - loop management from "pre - hospital, in - hospital, to post - hospital". Qiao Yuchen, the product manager of Zijing Zhikang, introduced the product development concept and current functions:
Before diagnosis, patients can register online and complete intelligent pre - consultations and generate structured medical records through conversations with AI robots. During diagnosis, doctors can view structured medical records on the workstation, saving some time for consultations and medical record writing. At the same time, AI doctors will provide suggestions for examinations and diagnoses, allowing real - life doctors to focus on key medical decisions. After diagnosis, the entire chain of medical data will be stored in the patient's health record. Subsequently, patients can also use services such as AI health consultations and interpretation of physical examination/inspection reports. The system will manage health indicators based on the timeline and provide health suggestions.
On June 30, 2025, the Zijing AI Hospital system was launched. In August, the system conducted internal tests of offline outpatient functions in departments such as general practice and respiratory medicine at the Tsinghua University Hospital.
Zijing Zhikang plans to conduct public tests of the system by the end of 2025. The test areas will expand from Beijing to more cities across the country, and the test scope will also cover hospitals of different levels and sizes, as well as more departments and scenarios.
In October 2025, five departments including the National Health Commission issued the "Implementation Opinions on Promoting and Regulating the Application and Development of 'Artificial Intelligence + Healthcare'". The Zijing AI Hospital is highly in line with the spirit of the document and is expected to effectively promote the empowerment of primary healthcare by artificial intelligence and improve the capacity and efficiency of medical services.
Investors' Views:
Li Wenjue, a partner at Xinglian Capital, said: "The Zijing AI Hospital represents the cutting - edge exploration of artificial intelligence in the medical field. It not only has original breakthroughs in technology but also creates a new growth paradigm on the medical supply side. It turns the concept of 'AI doctors' into reality, reshapes medical efficiency and fairness with agent technology, and makes high - quality medical services accessible. We believe that with the team's scientific research strength and industrialization capabilities, the Zijing AI Hospital will become an important force driving the transformation of smart healthcare infrastructure in China and even the world. We look forward to working with the team to jointly promote the birth of a new - era medical system featuring human - machine collaboration."
An investor from Inno Angel Fund said: "Inno has always been concerned about the application of large AI models in the medical field. Limited by the shortage of clinical medical record data, the cost of training large medical models has remained high. The Zijing AI Hospital generates medical record models based on clinical guidelines, providing sufficient data for medical agents and enabling their self - evolution. The diagnosis and treatment of more than 20 diseases have reached the level of chief physicians, and hospital tests have been completed, improving the accuracy of consultations and medical efficiency, allowing the public to enjoy higher - quality and higher - level medical services. Inno is optimistic about the Zijing team and hopes that it will quickly promote the public testing and popularization of the product and contribute to the construction of a 'Healthy China'."
Liu Xiaolei, a partner at Shangshi Capital, said: "Shangshi Capital is very concerned about the transformation brought by AI technology development to the medical industry. We have been following Zijing Zhikang since its establishment and are very glad to participate in this round of investment. The core highlight of the project lies in its 'virtual hospital + evolvable agent' technology path. The Zijing Zhikang team has broken through the data bottleneck, enabling AI doctors to self - iterate in a simulated environment with an accuracy rate of over 96% and seamless migration to real scenarios. Backed by Tsinghua's academic and clinical resources and benefiting from the national policy dividends of 'artificial intelligence + healthcare', the project has a scarce full - process closed - loop and cross - model architecture. It is expected to reshape medical supply, alleviate the shortage of primary doctors, and has both great social value and commercial potential."
Zhang Yu, the founding partner of Qingzhi Capital, said: "As the seed - round investor of Zijing Zhikang, Qingzhi Capital has participated in the team's transformation from scientific research to industry. We are full of confidence in the company's development. The entrepreneurial spirit of the Zijing Zhikang team to 'do difficult but right things', the profound scientific research accumulation of the core team in industry models, the precise positioning of the AI Agent Hospital, the open and inclusive industrial ecosystem, and the perseverance in solving the pain points of the medical industry made us choose to invest in the seed round without hesitation. We look forward to Zijing Zhikang's continuous in - depth exploration of technology, improvement of products and services, and becoming a new - force leading the transformation of the medical technology industry. Qingzhi Capital will also continue to provide all - round support as always."