»Lianying Intelligent« setzt nach Serie-A-Finanzierung von 1 Milliarde Yuan verstärkt auf mehrmodale medizinische KI-Systeme | Projektbericht

Text by | Hu Xiangyun

Edited by | Hai Ruojing

36Kr learned that United Imaging Intelligence, a subsidiary of United Imaging Group that operates independently and focuses on medical artificial intelligence technology, is planning to consider an independent listing . In June, United Imaging Intelligence announced the completion of a 1 billion yuan Series A financing round. The investors included E Fund, Shanghai Guotou Asset Management, and Shanghai United Investment, among others.

At the recently concluded 8th World Artificial Intelligence Conference (WAIC), Zhou Xiang, the co-founder and co-CEO of United Imaging Intelligence, further told 36Kr that this round of financing will, on the one hand, expand into multi-modal applications, integrating imaging, text, voice, and other modalities for use in scenarios such as operating rooms and wards. On the other hand, it will focus on the R & D and upgrading of large model agents and the hospital's digital and intelligent business systems.

The company has launched 12 product platforms and over 100 AI applications. It has obtained 13 Class III medical device certificates, 15 AI applications have passed the US FDA certification, and 31 AI applications have received the CE certification. Currently, United Imaging Intelligence's AI products have entered more than 4,000 medical institutions across the country, covering multiple medical scenarios such as hospital management, imaging-assisted diagnosis, surgical treatment, medical record writing, and scientific research applications.

Since the beginning of this year, with the enhanced capabilities of local deployment and model customization provided by general large models such as DeepSeek, medical institutions at all levels have included "AI in hospitals" as a key focus. Among them, large models in the medical vertical domain have become an important part. Compared with general large models, they are more suitable for handling the complex and diverse medical needs within hospitals.

Relying on United Imaging Group's accumulation in the fields of medical imaging and medical AI, United Imaging Intelligence has independently developed the "Yuanzhi" medical large model, which integrates multi-modal large models such as text, imaging, vision, voice, and hybrid models. While absorbing the natural language and long-text processing capabilities of general large models like DeepSeek, it can also give birth to self-evolving, multi-modal, and adaptive medical agents according to different medical scenario requirements through vertical domain specialization and private knowledge fine-tuning, as well as the integration of the capabilities of different modal large models.

Take its newly launched "Radiology Agent" as an example. It is a chest multi-disease screening agent. After a single chest CT scan, it can automatically identify 73 types of chest abnormalities, such as lung nodules, pneumonia, pulmonary hypertension, and coronary artery calcification.

This "all-in-one" effect of multi-disease screening is fundamentally different from the single-disease AI products commonly developed in the past and is not simply a stack of multiple single-disease small models. "In the past, due to technological limitations, CNN (Convolutional Neural Network) models were small, and only one model could be developed for one disease. For example, if you were looking at lung nodules alone, you couldn't analyze both lung nodules and coronary arteries simultaneously. The products were just piled up, and not many algorithms could be run. Now, with the Transformer + CNN model, when the data volume is large enough, the model can analyze the relationships between different diseases, thus achieving'multi-disease screening from a single scan'."

In addition, the medical agent products currently deployed by United Imaging Intelligence also include surgical agents, interventional doctor agents, electronic medical record agents, customer service agents, and quality control management agents.

Zhou Xiang introduced that three factors - medical value, technological maturity, and business opportunities - are indispensable when determining whether a medical scenario can support the development of an agent. Based on these considerations, what United Imaging Intelligence currently wants to do is to "achieve digitalization across the entire hospital." For example, the digital upgrade of the hospital's business systems and the information systems in scenarios such as operating rooms and wards may all be potential directions.

The market space for each of these scenarios may not be very large, but as the number of developed products increases and the marginal cost decreases, they may bring new growth opportunities.

For example, for tertiary hospitals, using AI to conduct special examination items (such as adolescent scoliosis screening and CT bone density monitoring) or high-end examination items (such as cardiac CMR and CMR cardiac function parameter measurement) on existing equipment can enable the hospital to achieve stronger differentiated competition.

Data from a Wuhan tertiary hospital that has cooperated with United Imaging Intelligence shows that for the DR full-spine scan examination project, in the past, due to the manual measurement of all parameters by doctors, large-scale implementation was not possible. However, after the introduction of AI, radiologists can use AI to automatically obtain accurate parameter results, which not only supports clinical diagnosis and surgical implementation but also improves the examination efficiency. Statistics show that after the introduction of AI, the hospital conducted more than 5,000 DR full-spine scan examinations from February to December 2024.

Meanwhile, for primary medical institutions, due to the lack of professional capabilities, it is difficult to carry out some important examination projects. However, the introduction of AI technology can effectively make up for the "soft power" deficiencies of primary hospitals. New examination projects can be carried out without adding new equipment or applying for new charging catalogs, while improving the hospital's diagnosis and treatment level.

Take a second-tier county-level hospital in Zhejiang as an example. In the past, the hospital's ability to perform vascular CTA examinations was relatively weak, and it had to transfer patients who needed coronary CTA and head and neck CTA examinations to another hospital in the city. After introducing relevant hardware equipment and coronary AI-assisted diagnosis software products in 2024, the hospital can independently carry out coronary CTA and other vascular enhancement examinations, and the number of examinations exceeded 1,000 cases that year.

"Hospitals are willing to pay for products that can solve practical problems, such as those that help hospitals improve their medical service levels and attract or retain patients. Hospitals are willing to buy such products," Zhou Xiang said.

In addition to clinical products, AI-enabled scientific research is also an area where United Imaging Intelligence is currently focusing its layout. For example, the company has joined hands with universities and hospitals such as ShanghaiTech University, Children's Hospital of Fudan University, and Xiamen Children's Hospital, and has been approved for two projects in the Chinese Brain Project targeting brain research on infants aged 0 - 6 and school-age children aged 6 - 18, namely "Research on the mapping of dynamic brain atlases of infants and the development mechanisms of language and social emotions based on new-generation artificial intelligence technology" and "Research on the brain and intelligence development cohort of Chinese school-age children," thus exploring the "uncharted territory" of brain science with AI.

It is understood that in terms of the structure of paying customers, most of the revenue from United Imaging Intelligence's scientific research platform comes from hospitals above the middle level, while the clinical application products cover a wider range, including top-tier tertiary hospitals, county-level hospitals, and health centers.