The editor of Downcodes reports: Alipay’s large medical model is coming in force, outperforming GPT-4 in the Chinese and English medical examinations, and has been implemented in first-line hospitals in Jiangsu, Zhejiang and Shanghai! This large multi-modal medical model not only achieves outstanding results in various evaluations, but also shows amazing accuracy in image recognition. More importantly, behind it is Alipay's huge medical and health service system accumulated over ten years and in-depth cooperation with many medical institutions, which provides strong impetus for the future development of AI medical care.
Alipay's large medical model has been unveiled! In the Chinese and English exams, its performance surpassed GPT-4 and has been implemented in first-line hospitals in Jiangsu, Zhejiang and Shanghai.
This large medical model not only exceeded the level of GPT-4 in the Chinese and English medical examinations and benchmark tests, but also achieved first place in the A list and second place in the B list in the Chinese medical LLM evaluation list promptCBLUE. What’s even more surprising is that its accuracy rate in identifying images such as reports, medicines, hair, etc. is over 90%.
At the WAIC site, Alipay teamed up with 20 institutions including People's Medical Publishing House, Peking University School of Medicine, and Zhejiang Provincial Health Commission to jointly launch an AI medical co-construction plan.
Alipay's large medical model is different from large models in other fields. It puts forward higher requirements in terms of multi-modality, security and professionalism. It is based on the Bailing large model and has native multi-modal capabilities of seeing, listening, speaking, and drawing. It can directly understand and train multi-modal data such as audio, video, images, and text. On this basis, the model also adds tens of billions of Chinese and English graphics and texts including reports, images, and medicines, hundreds of billions of medical text corpus, and tens of millions of high-quality medical knowledge maps.
At present, this large multi-modal medical model can not only provide intelligent question and answer, medical record structuring and retrieval, auxiliary diagnosis, but also identify drugs and hundreds of medical reports, conduct hair health testing, etc. These capabilities can be directly embedded in the medical links of hospitals and other institutions and applied quickly.
During the research and development phase, Alipay cooperated with hundreds of professional medical teams and experts to label data to ensure the authority and credibility of the information. At the same time, together with Shanghai Renji Hospital, the first Chinese medical specialty question and answer reasoning data set RJUA-QA was launched. During the model testing phase, users, patients, and doctors participate in question and answer feedback to ensure the practicality and accuracy of the model.
In terms of security, Alipay has taken a number of measures to ensure technical reliability and data privacy security. Ant itself has been deeply involved in the security field for many years. This time, it has launched a "medical trusted all-in-one machine + trusted cloud" solution for the industry, which realizes the integration of training and promotion and the collaborative optimization and acceleration of domestic computing power software and hardware to solve the problem of hospital computing. solve the problem of insufficient power; deploy dense state reasoning on public clouds and private clouds to solve data privacy and security issues.
On the basis of this large model, Alipay also released full-modal digital humans, intelligent agent-related products and industry solutions. In fact, Alipay has already started deploying AI in medical scenarios as early as last year. In November last year, the Zhejiang Health Commission applied the "AI Medical Assistant" solution opened by Alipay to create the country's first digital health person "Anclin" to provide Zhejiang residents with cloud medical companionship, health consultation and other services.
Earlier, Alipay supported the country's first remote registration payment in 2014. After ten years of development, the platform has served more than 600 million people seeking medical treatment, becoming the largest third-party medical insurance online service platform and one-stop medical and health service platform in China. More than 3,600 hospitals in more than 300 cities across the country provide one-stop digital medical services on Alipay.
As one of the three core application areas of Ant, medical care has naturally attracted much attention. The three major areas in which Ant's large model is implemented are medical care, finance and life, which are expressed in the form of professional intelligence agents: AI medical assistant, AI financial assistant and Alipay intelligent assistant.
At WAIC, Alipay intelligent assistant made its debut and was selected as this year’s treasure. You can wake it up by scrolling down on the Alipay homepage. You can complete services such as errands, consultations, and taxi rides through conversations, and you can call thousands of mini programs at any time. For example, if you want to order a cup of Starbucks' tall iced latte, you only need to give an instruction, and the AI will automatically place the order. After the user confirms and pays, he can go to a nearby offline store to pick up the coffee.
Jing Xiandong, chairman and CEO of Ant Group, said that professional agents are an effective path for large-scale models to be implemented in rigorous industries. Intelligent cognitive decision-making technology and platform are key technologies that support the three major intelligent agents, and have also won the important award of this conference "Excellence in Artificial Intelligence Award (SAIL) TOP30". This technology was jointly researched by Ant Group and Zhejiang University for seven years. It has greatly improved the professional capabilities, accuracy and self-learning capabilities of artificial intelligence in serving rigorous industries, making AI and large model decision-making smarter and more reliable.
Up to now, the project technology has been widely used in medical and livelihood, finance, government and other industries, serving more than 2,600 hospitals and institutions such as Shanghai Renji Hospital, Shanghai No. 1 Hospital, Zhejiang Provincial Health Commission, etc. Ten financial institutions have promoted the large-scale implementation of artificial intelligence. The project has also been authorized for more than 100 invention patents, and some of its results have won the first prize of the China Electronics Society's Science and Technology Progress Award in 2022 and the first prize of the Wu Wenjun Artificial Intelligence Science and Technology Progress Award in 2023.
Alipay's large medical model not only allows people to see the broad application prospects of AI in the medical field, but also demonstrates its powerful capabilities in improving medical efficiency and ensuring data security. In the future, with the continuous advancement of technology and the deepening of its application, AI medical large models are expected to bring more convenient and smarter medical service experiences to more people.
All in all, the emergence of Alipay's large medical model marks an important step for AI in the medical field. Its powerful performance and wide application prospects are worth looking forward to. The editor of Downcodes will continue to pay attention to its development and bring more relevant reports to readers.