Recently, a research achievement jointly completed by the Department of Bioinformatics and Biostatistics, School of Life Sciences and Biotechnology, and the Digital Medical Technology Center, Institute of Translational Medicine at Shanghai Jiao Tong University was successfully published in Advanced Science. The three co-first authors are Deng Luojia, a Zhiyuan Ph.D. student from the School of Life Sciences and Biotechnology, Wu Yijie, a Ph.D. student from the same school, and Ren Yongyong, an experimentalist and doctor from the Institute of Translational Medicine. Lv Hui, a distinguished professor at the School of Life Sciences and Biotechnology, served as the corresponding author. The team proposed and constructed a truly intelligent system that can "conduct scientific research on its own" - DREAM (Data-dRiven self-Evolving Autonomous systeM). From this point forward, AI is no longer just a tool but a collaborator in exploring the unknown world, sketching the dreams of the next-generation scientific research civilization at the boundaries of science.

Against the backdrop of the increasingly prominent contradiction between massive amounts of data and limited human resources, DREAM is the world's first system capable of independently completing the entire scientific exploration process. From formulating scientific questions, writing analysis procedures, performing computations to validating results, the whole process requires no human intervention. Through modular intelligent agent collaboration, the system automatically completes the four key scientific research links of "questioning - coding - configuring - evaluating," completely freeing researchers from repetitive labor.

When applied to the classic Framingham Heart Study data, the DREAM system demonstrated an efficiency over 10,000 times that of ordinary researchers. It can even autonomously discover important medical associations that can be verified decades in advance, representing a leap from "auxiliary tool" to "autonomous scientific research entity" for biomedical AI.

This breakthrough not only marks a new stage where AI transitions from a research assistant to a research subject but also provides a paradigm reference for constructing sustainably evolving AI scientific research systems. It offers a powerful technological path for promoting large-scale, automated, and autonomous scientific discoveries, enabling scientists to freely explore the unknown realms of science.

Paper Link: https://onlinelibrary.wiley.com/doi/10.1002/advs.202417066