Artificial intelligence (AI) is already playing an increasing role in healthcare in China. At Xuanwu Hospital in Beijing, a radiologist is processing a case of head and neck computerized tomography angiography with the help of an artificial intelligence diagnostic doctor system, which takes only about three minutes to confirm the diagnosis. The AI doctor system at the Radiology Department of the public 3A-grade hospital does most of the work of reading the imaging as well as generating reports in advance, a far cry from the time when the radiologist needed more than 100 mouse clicks and about 30 minutes to complete the diagnosis. About 1,100 kilometers away in Xian, Shaanxi province, at a small primary hospital, a doctor is conducting an ultrasound test on a female patient to scan for possible breast cancer risk. As the doctor scans the patient, the AI doctor system installed there detects lesions simultaneously, greatly assisting in completing and improving the results.
“AI technologies are exerting great influence on the healthcare industry, both in preventive care and treatment,” said Anne Ma, CEO of Shukun Technology, the provider of the AI doctor system. “With the latest evolution of generative AI, it is possible for us to build a personalized AI doctor for every human being. We are very passionate and inspired by the vision.” China, which has several medical AI companies like Shukun Technology, is speeding along in the AI industry’s fast lane.
According to data from the Ministry of Industry and Information Technology (MIIT), the scale of China’s core AI industry has reached CNY500 billion, and the number of related enterprises has exceeded 4,400. China’s medical large model industry is set to witness a period of explosive growth between 2023 and 2027, with its market size forecast to reach CNY22.25 billion by 2030, a 36% growth from the estimated level in 2029, a report on the industry released last year by Beijing-based think tank EO Intelligence showed. By October, there were 238 large models (LMs) in China, including nearly 50 medical LMs, covering areas such as patient inquiries, doctor assistance, drug research and development, and medical popular science.
Compared to conventional medical AI, medical LM is more like a human brain, which is capable of understanding human language, completing logical deductions and generating final results. Currently, there is a wide range of applications for medical LMs. Researchers from Fudan University in Shanghai and the University of Massachusetts in the United States recently had their latest medical LMs take the U.S. Medical Licensing Exam, and the results showed AI surpassing 70% of medical students. Doctors are maintaining an open attitude to the proliferation of medical LMs. In the field of oncology, for example, most oncologists hold a positive attitude toward the application of AI, said a recent report conducted among healthcare professionals by Dingxiangyuan, an online health information services provider. Doctors, it said, are constantly understanding its true value through clinical practice and applications.
The demand for large-scale modeling technology from enterprises has risen accordingly. According to U.S. market consultancy Gartner’s research earlier this year, over 60% of Chinese enterprises plan to deploy generative AI within the next 12 to 24 months, and healthcare is one of the most important application scenarios. “Speaking of AI, we physicians from the Imaging Department may benefit the most. AI is able to replace our preliminary work, such as reading images and making primary assessments. However, we also still need to read images and make diagnoses by ourselves,” said a physician surnamed Qin, who works at Beijing's Chaoyang Hospital.
Through big data and deep learning, medical LMs cut intermediate links, recommend diagnosis and treatment for doctors and patients, and enhance work efficiency to a large extent, said Zhang Shule, a columnist at people.cn. However, because of the complexity of many diseases, medical LMs are often not able to complete the entire diagnosis and treatment procedure. “This pain point requires sufficient and vertically segmented big data accumulated for different cases, to provide deep learning models for diagnosis and treatment reference, in order to minimize the misdiagnosis rate of doctors. However, such a large amount of data cannot be accumulated solely by one city or province, and requires nationwide data exchange and reference to foreign cases, which is somehow difficult to achieve,” Zhang said. Experts also said that the serious nature of healthcare, a lack of interconnectivity in data, and the industry’s zero fault tolerance make the commercialization of medical LMs difficult, the China Daily reports.
China is narrowing the artificial intelligence (AI) gap with the U.S. through rapid progress in deploying applications and state-backed adoption of the technology, despite the lack of access to advanced chips, according to industry experts and analysts. Chinese tech firms have rushed to create their own large language models (LLMs) – the underlying technology behind generative AI technologies like ChatGPT – with many even claiming to match or exceed their U.S. counterparts, all amid tighter U.S. restrictions on advanced chips considered critical to the training of AI systems. For example, Shengshu AI, a little-known start-up based in Beijing, launched its text-to-video tool last week, becoming the latest local firm to offer a Sora-style service for unlimited public use, after Kuaishou and Zhipu AI. The tool, called Vidu, is able to generate clips from Chinese and English text prompts. While text-to-video was pioneered by Sora, the three Chinese tech firms have been able to put their AI video tools in the hands of global users. In comparison, San Francisco-based start-up OpenAI, which was the first to demonstrate the function, has yet to make its tools widely available. Chinese firms are also contributing to global AI development by launching open source LLMs so anyone can build their own AI systems, the South China Morning Post adds.
