Under the background of deep integration between AI and healthcare, Quark has officially released the first domestic comprehensive test set for all stages of physician qualification exams. This test set not only provides comprehensive support for the physician qualification exam but also lays a solid foundation for the development of AI models in the medical field.
According to the information, Quark's health large model has successfully passed the written examination evaluation of the chief physicians in 12 core medical disciplines in China, and has published the technical report "QuarkMed Technical Report", which details the implementation details of its "chief physician-level" capabilities. The release of this test set fills the gap in high-timeliness and wide-coverage physician examination test sets in China, marking another important advancement in AI medical applications.
This test set covers four different levels of exams: practicing physicians, associate chief physicians, deputy chief physicians, and chief physicians. It includes approximately 7,600 questions from 12 core disciplines such as general internal medicine, surgery, gynecology, and pediatrics. The real questions are all taken from the latest 2024 exam content. The design of the test set adopts an effective stratified strategy, making the hierarchy of exams at different levels clear. The primary level exams mainly consist of single-choice questions and memory-based questions, while as the level increases, the proportion of multiple-choice questions and case analysis questions gradually increases, along with higher requirements for model reasoning ability.
Wang Qiangyu, CEO of Yisheng Jiankang, stated that the release of this test set will save a lot of time and effort for start-up teams. When developing models related to gynecology, the team encountered difficulties in evaluating model performance. Wang Qiangyu mentioned that general medical benchmarks have poor adaptability in specific scenarios. With Quark's professional test set, the team can focus more on algorithm optimization and scenario adaptation.
Through this innovative measure, Quark not only promotes the standardization of physician exams, but also provides important support for further application and development of AI in the medical field.
