The AI craze is sweeping across all walks of life, and the medical industry is no exception. At the same time, the emergence of AI is also expected to bring disruptive reforms to the medical industry. Currently, all parties are focusing on how AI can be applied in specific medical scenarios.
On the morning of September 6, at the 2024 REAL Technology Conference hosted by Jiemian News, at a roundtable discussion with the theme of "AI + Medical: Hot Layout and Pragmatic Thinking", Ma Rui, partner of Fengrui Capital, and President of the Life Sciences Division of Shenzhen Technology Manager Li Xiaobing, Gao Yushi, Vice President of Technology of Easy Group, Xiang Lei, CTO (Chief Technology Officer) of Shenzhen Zhitong Medical, and Lou Yang, Managing Director of Light Source Capital, had a lively discussion on this topic.
Gao Yushi, Vice President of Technology of Easy Group, shared that Easy Group·Easy Health released the large medical and health model Dr.GPT in May 2023, and made a major upgrade at the end of 2023 with the expansion of applications and the deepening of scenarios. , released seven major applications based on the "Easy Doctor Dr.GPT" model, covering all scenarios of health management services and meeting the specific needs of different application scenarios and user groups. In addition to providing users with comprehensive health management support, it also expands the capabilities of medical diagnosis and treatment technology, providing a more efficient and comprehensive perspective for medical decision-making.
In addition, one of the biggest controversies in AI+ medical applications is data security. Xiang Lei, CTO of Shenzhen Zhitong Medical, said that the issue of data privacy is the biggest concern for hospitals or doctors. Currently, the method of moving data to the cloud is internationally acceptable. For example, Amazon Cloud has passed the information privacy protection mechanism. Third-party companies use Amazon Cloud to provide services to hospitals, and hospitals recognize this method. Data control in China is stricter, and the hospital will require that all data must be on the client and cannot be uploaded to the cloud.
Based on different data usage methods, Xiang Lei said that Shenzhen Zhitongyi's business model in the cloud is to charge on a case-by-case basis, while in China, due to localized deployment, a one-time payment method is adopted, "different solutions are provided according to the specific needs of the client." The solution can meet the needs of customers and the application and commercial sales of products in specific scenarios.”
Xiang Lei said that the model has developed into the 2.0 era. Compared with the 1.0 era, which only required a small amount of data for small application scenarios or developed exclusive models for customers, the 2.0 era can obtain a large amount of data, and it is expected to support all through a unified model. Hospital.
At this stage, Shenzhi Touyi uses a universal model that can process various imaging modalities. It has developed a model for different departments that can process all modalities simultaneously. Xiang Lei said that this kind of processing effect is better than a single model. “This is also the result of the general basic model plus a large amount of data training. In addition to the existing scenarios, we also found that using multi-modal data Better results can be achieved.”
At present, in addition to the application of AI on patients and hospitals, AI pharmaceuticals is also a mainstream application direction. Li Xiaobing, general manager of the Life Sciences Division of Shenzhen Technology, said that the current mainstream business models include AI+software, AI+CRO, and AI+Biotech, and Shenzhen Technology has a presence in all three aspects.
Li Xiaobing said, "Shenzhen Technology currently ranks first in the industry's market share in some physical computing tools on the AI SaaS side; in the AI+CRO model, it has cooperated with leading domestic pharmaceutical manufacturers including Fosun and Dongyangguang Pharmaceutical. In cooperation, Shenzhen Technology provides AI+ design solutions, and the other party provides verification and joint research and development models; in terms of AI+Biotech, Shenzhen Technology is also trying to incubate some drug pipelines internally. Among these three directions, the AI+SaaS side is. Focus on the direction of investment.”
Ma Rui, a partner at Fengrui Capital, shared the factors for the long-term development of AI+ healthcare from a capital perspective. Ma Rui said that returning to investment logic, the most important thing in the long run is data. Currently, an important investment direction of Fengrui Capital is the digitization of biological systems and biological processes. Whether it is calculated, measured, or sensed, increasing data is its long-term optimistic direction. However, Ma Rui believes that the most important thing is to understand AI, "How to use AI in the biological field, how to combine physics and AI, and how to use large models as the base. In fact, you don't need to do too many experiments to get your results." The desired result is what we are seeing now.”
Talking about the future, Xiang Lei hopes to be more closely integrated with doctors. He hopes that doctors will use AI more as a tool to help them make decisions and improve the accuracy and efficiency of diagnosis. It used to take half an hour to make a diagnostic report. , now it may be done in 5 minutes or 3 minutes, which will ultimately benefit the patient.
Gao Yushi believes that large AI models provide strong technical support for the realization of 4P medical theory. Large models can integrate medical data to build disease prediction models, analyze gene sequences, medical images and population health data, and support predictive and preventive medicine. At the individual level, individual multi-modal data can be deeply analyzed to formulate personalized treatment plans and make real-time adjustments to promote the development of personalized medicine. In addition, smart medical assistants can provide patients with convenient services and health management tools, improve patient participation, and are expected to promote changes in medical models and bring greater benefits to human health.
Li Xiaobing expressed his expectations for the research and development of new drugs. He believes that in the next 2-3 years, AI will play a huge role in breakthroughs in some points, such as early development of molecular design, molecular evaluation, molecule generation, and a series of molecular-level research. This aspect will help scientists provide higher-throughput design solutions or more creativity. However, drug research and development has many levels, from the molecular level to the cellular level, to organs, and then to the human body, which requires AI to achieve a certain level of technology accumulation.
"Just like drugs behave differently in cells, small animals, and people, we need a process of AI emergence, just like the process of life emergence, from molecules to cells, to organs, and then to people." Li Xiaobing said
Ma Rui also expressed his expectations for the future of AI + medical care. He believes that driven by AI, the understanding of biology will become deeper and deeper, and more and more things can be done with biotechnology as the underlying layer, such as biomedicine and biomanufacturing. , medical equipment, bio-agriculture, etc., this can be used as the underlying energy to radiate. Ma Rui believes that in 10 years, there will be many opportunities to emerge in the AI+biology and AI+medical tracks.