Tsinghua University-incubated Start-up Expands Virtual Hospital Testing with 'AI Doctors'
Tsinghua University-incubated start-up, Tairex, is conducting internal tests of its virtual hospital platform powered by AI. The platform, Agent Hospital, features 42 AI doctors across 21 departments and is set for public testing in 2025. China's AI industry is projected to invest over US$1.4 trillion in developing AI technology in the next six years.
Tairex, established in September under the university's Institute for AI Industry Research (AIR), has initiated internal trials of its platform. This innovative system boasts 42 AI doctors across 21 departments, including emergency, respiratory, paediatrics, and cardiology, as detailed in a recent AIR blog post.
The platform, dubbed Agent Hospital, is set to undergo public testing in the first quarter of 2025, with plans to make it accessible to the general public by the first half of the following year. Tairex, headquartered in Wuxi, has already generated over 500,000 AI patients from diverse demographics and medical conditions for its virtual hospital platform. The company encourages both the public and medical professionals to engage with the platform and participate in testing.
China's strides in integrating AI into healthcare and other sectors are closing the technological gap with the United States, leveraging the country's vast market, rich data resources, and expanding industrial foundation. Forecasts suggest that China's AI industry could invest over 10 trillion yuan (US$1.4 trillion) in advancing this technology over the next six years, according to Chen Liang, chairman of China International Capital Corp, a state-backed investment entity.
Tsinghua University's AIR introduced the concept of an AI-based medical platform earlier this year in a paper titled "Agent Hospital: A Simulacrum of Hospital with Evolvable Medical Agents." The platform employs agents representing patients and medical professionals, created using a large language model (LLM) akin to ChatGPT, to simulate various medical processes such as registration, consultation, and examination. The objective is to train these doctor-agents to handle tasks like diagnosis and treatment recommendations.
The doctor-agents, generated using GPT-3.5, achieved impressive accuracy rates of 88%, 95.6%, and 77.6% in examination, diagnosis, and treatment tasks, respectively. Notably, each AI doctor on the platform can efficiently manage tens of thousands of cases within days, a feat that would take a human doctor several years to accomplish. This innovative approach eliminates the need for laborious manual data labelling, as highlighted in the AIR paper.
In a similar vein, the Shanghai East Hospital recently introduced an LLM named MedGo, based on Alibaba Group Holding's Qwen2-72B model, to aid doctors in diagnosis and treatment. This LLM was trained on an extensive medical data set and textbooks, showcasing the growing integration of AI in the healthcare sector. Meanwhile, in Hong Kong, the Chinese Academy of Sciences' AI research centre launched a medical AI model named Cares Copilot, utilised by hospitals including the Prince of Wales Hospital and the First Affiliated Hospital of Sun Yat-sen University in Guangzhou.
Tsinghua University-incubated start-up, Tairex, is conducting internal tests of its virtual hospital platform powered by AI.
The platform, Agent Hospital, features 42 AI doctors across 21 departments and is set for public testing in 2025.
China's AI industry is projected to invest over US$1.4 trillion in developing AI technology in the next six years.
Source: SCMP