Summary of Key Points
Hangzhou Shulan Medical Group plans to complete the Liangzhu AI Hospital by the end of 2027. This is not merely about adding AI tools to a traditional hospital; rather, it aims to reconstruct medical services through computational medicine (AI + life science modeling), shifting from a focus on treating diseases after they occur to predicting and preventing them in advance, similar to the AI-driven healthcare model at the Mayo Clinic in the United States. However, as a new concept, the AI Hospital faces numerous challenges regarding data sharing, organizational structure, and space design. Shulan is still in the process of exploring these issues, with the goal of becoming a lifelong health manager for patients, ultimately achieving precision medicine tailored to each individual.
1. The AI Hospital Is Not Just About Adding AI Tools
Many people think an AI Hospital involves deploying some guidance robots and using AI to analyze medical images, but it’s much more than that. According to the “International AI Hospital Consortium Consensus,” an AI Hospital integrates AI deeply into every aspect of patient care, connecting both online and offline services to provide continuous and superior care.
How does this “deep reconstruction” happen? Here’s an example:
- Changes in Space: In the past, hospital outpatient halls had to accommodate large numbers of people waiting in line for registration. Now, AI handles appointments and triage in advance, reducing the need for large halls. Areas dedicated to health management (such as check-ups and preventive consultations) will become more prominent, and laboratory facilities can be more flexibly adapted (e.g., without the need for major renovations when equipment is updated).
- Changes in Processes: The traditional medical process was “register → wait in line → see a doctor → get medication.” With AI, patients are first triaged (for example, if you say you have a headache, AI determines whether you should see a neurologist or an internist). During the consultation, AI helps doctors with patient records and analysis of test results, and after the visit, it tracks your recovery and provides personalized health recommendations.
- Changes in Goals: Traditional hospitals focus on treating existing illnesses, while AI Hospitals aim at preventing them. For instance, they could predict several years in advance whether you are at risk of developing diabetes and then help you adjust your lifestyle to avoid the disease.
2. Learning from the Mayo Clinic: A Leader in AI Healthcare
Why does Shulan want to follow the Mayo Clinic’s approach? Because the Mayo Clinic is a global benchmark for AI healthcare:
- Ample Data: The Mayo Clinic has 26 PB of clinical data, equivalent to 26,000 terabytes of storage space, including 3 billion test results, 1.6 billion patient records, and 6 billion images. It also collaborates with other hospitals to analyze data from 15.2 million patients.
- Advanced AI Applications: This year, the Mayo Clinic partnered with Microsoft to develop a large medical model that combines its clinical expertise with Microsoft’s AI technology to solve specific healthcare problems. As early as 2019, the Mayo Clinic transitioned from a “pipeline-based” model (where you come in and we treat you) to a “platform-based” model (using AI and data to serve more people).
Shulan aims to follow this path: providing comprehensive in-hospital care while also managing patients’ long-term health outside the hospital and using real data to validate AI technologies. For example, their Mahong Baihan Hospital has turned its AI clinical trial tools into an independent company, indicating that top hospitals are commercializing their AI capabilities.
3. The Liangzhu AI Hospital as a Testing Ground
Shulan has made significant preparations:
- Two Sets of Operating Systems: One set manages internal hospital operations (such as doctor collaboration and equipment management), and the other focuses on patient care (lifelong health management). Their “Aladdin System” allows patients to register, consult doctors, and view reports online, while also generating personalized health profiles. The “Dr.Shu AI Intelligence Agent” provides services throughout the entire medical process, from pre-consultation to post-treatment assistance.
- Innovative Space Design: The outpatient area has been significantly reorganized (e.g., by reducing waiting areas), but more radical changes are needed for expensive spaces like operating rooms due to cost considerations.
- Organizational Challenges: How do teams in an AI Hospital collaborate effectively? In the past, multidisciplinary consultations involved one department treating the patient first and then seeking help from other departments if necessary. Now, AI and interdisciplinary doctors may work together from the beginning. The hospital has AI doctor teams that regularly evaluate the effectiveness of AI in treating cases to help doctors adapt to these tools.
However, Shulan is cautious: they avoid investing heavily in expensive resources like GPU servers but ensure they remain responsive to emerging AI technologies to stay competitive.
4. Data Is the Fuel for AI Healthcare, but Challenges Remain
The core of AI healthcare is computational medicine, which involves creating a “digital twin” for each patient to simulate the effects of treatments and predict risks. This requires complete, standardized, and shareable data. Current challenges include:
- Data Fragmentation: Health records are often stored in different places (e.g., one hospital’s lab results may be in another hospital, and wearable device data is on mobile devices), making it difficult to integrate them.
- Inconsistent Standards: Medical record formats vary across hospitals, making it hard for machines to process the data uniformly. For example, terms like “hypertension” are differently defined, hindering AI analysis.
- Data Sharing: Hospital data is often considered private, making cross-institutional and cross-regional sharing challenging, with privacy concerns (e.g., data anonymization and preventing patient information leaks) needing to be addressed.
Shulan has been actively working on these issues: in 2015, it initiated the OMAHA Alliance to promote standardization of medical data, and in 2024, it established a key laboratory for computational medicine in Zhejiang Province. However, these problems cannot be solved by a single hospital; they require collaboration from the government and the industry to establish clear guidelines.
5. A Bright Future: Precision Medicine Is Within Reach
Zheng Jie believes that the era of personalized medicine (one drug and one treatment plan for each patient) is indeed coming. This is not just empty talk. For example, with digital twins, doctors can simulate the effects of different medications in a virtual environment to find the most suitable one for you, or they can predict your risk of developing certain diseases and intervene early.
Still, there are many challenges to overcome. For instance, digital twins are currently being tested on a limited scale (first managing patient history data before building more comprehensive virtual models), and the balance between organizational structure and technology investment needs to be carefully balanced. Nevertheless, Zheng Jie is confident that the cycle of using real data to improve AI and then using AI to serve patients will accelerate, creating a virtuous cycle that ultimately leads to precision medicine.
In summary, AI Hospitals represent the future of healthcare, but there is still a long way to go. Shulan’s Liangzhu project is a bold attempt that could transform the entire industry if successful.