The Alipay Medical Large Model has been launched! It has outperformed GPT-4 in both Chinese and English exams and has been implemented in first-tier hospitals in Jiangsu, Zhejiang, and Shanghai.
This medical large model not only exceeds the level of GPT-4 in Chinese and English medical exams and benchmark tests, but also achieved first place on the A list and second place on the B list in the Chinese medical LLM evaluation list promptCBLUE. Even more surprisingly, its accuracy in identifying images such as reports, medicines, and hair has reached over 90%.
At the WAIC, Alipay, in collaboration with People's Sanitation Press, Peking University School of Medicine, Zhejiang Provincial Health Commission, and other 20 organizations, initiated the AI Medical Co-construction Program.
Alipay's medical large model, different from other large models in other fields, puts forward higher requirements in terms of multimodality, security, and professionalism. Based on the Bailing Large Model, it has native multimodal capabilities of seeing, listening, speaking, and painting, and can directly understand and train multimodal data such as audio, video, images, and text. On this basis, the model also adds billions of Chinese and English textual and graphic data, including reports, images, and medicines, as well as tens of billions of medical text corpus and millions of high-quality medical knowledge graph data.
Currently, this multimodal medical large model can not only provide intelligent question and answer, case structure and retrieval, and assist in diagnosis, but also identify medicines and more than a hundred types of medical reports, conduct hair health tests, and more. These capabilities can be directly embedded into the medical processes of hospitals and other institutions for quick application.
In the research and development stage, Alipay collaborated with hundreds of professional medical teams and experts to label data, ensuring the authority and credibility of information. At the same time, it launched the first Chinese medical specialty question and answer reasoning dataset RJUA-QA in collaboration with Shanghai Renji Hospital. During the model testing stage, user, patient, and physician feedback was involved to ensure the practicality and accuracy of the model.
In terms of security, Alipay has taken multiple measures to ensure the reliability of technology and data privacy. Ant Group has a deep focus on the security field for many years, and this time it has launched the "Medical Trustable Integrated Machine + Trustable Cloud" solution for the industry, achieving integrated training and inference and coordinated optimization of domestic computing hardware and software, solving the problem of insufficient computing power in hospitals; through public and private cloud deployment of secret inference, solving the problem of data privacy and security.
On the basis of this large model, Alipay has also released full-modal digital human, intelligent body-related products, and industry solutions. In fact, Alipay has already started the layout of AI in the medical field last year. In November last year, the Zhejiang Provincial Health Commission applied the "AI Medical Assistant" solution opened by Alipay to create the first digital health person "Anzhen'er" in Zhejiang, providing cloud companionship and health consultation services for Zhejiang residents.
Even earlier, Alipay supported the first remote registration and payment in the country in 2014. After ten years of development, the platform has served over 600 million people seeking medical care, becoming the largest third-party online health insurance service platform and one-stop medical 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 fields of Ant, medical care naturally attracts a great deal of attention. The three fields where the Ant Large Model is landing are medical care, finance, and life, with the form of professional intelligent bodies: AI medical assistant, AI financial assistant, and Alipay intelligent assistant.
At the WAIC, the Alipay intelligent assistant made its first appearance and was selected as this year's featured exhibit. By pulling down the Alipay homepage, you can activate it, complete services such as consultation, ride-hailing, and calling thousands of mini-programs through dialogue. For example, to order a medium cup ice latte from Starbucks, simply give the command, and the AI will automatically place the order. After the user confirms and pays, they can pick up the coffee from the nearest offline store.
Wellin Wang, Chairman and CEO of Ant Group, said that professional intelligent bodies are an effective path for the strict implementation of large models in industries. The intelligent cognitive decision technology and platform are the key technologies to support the three intelligent bodies and have also won the important award "Excellent Artificial Intelligence Award (SAIL) TOP30" at this conference. This technology has been jointly researched and developed by Ant Group and Zhejiang University for seven years, significantly improving the professional capabilities, accuracy, and self-learning ability of artificial intelligence services to rigorous industries, making the decision-making of AI and large models smarter and more reliable.
So far, the project technology has been widely applied in industries such as medical and people's livelihood, finance, and government affairs, serving more than 2,600 hospitals and institutions including Shanghai Renji Hospital, Shanghai No. 1 Hospital, Zhejiang Provincial Health Commission, and dozens of financial institutions, promoting the large-scale implementation of artificial intelligence. The project has also been authorized more than 100 invention patents, and some of the results have won awards such as the 2022 China Electronics Society Science and Technology Progress Award and the 2023 Wu Wenjun Artificial Intelligence Science and Technology Progress Award.
The Alipay Medical Large Model not only shows the wide application prospects of AI in the medical field but also demonstrates its powerful capabilities in improving medical efficiency and ensuring data security. With the continuous advancement of technology and the deepening of applications, AI medical large models are expected to bring more convenient and intelligent medical service experiences to more people in the future.