The world's first generative artificial intelligence (AI) radiology system embedded in clinical processes, developed by Northwestern University School of Medicine, can identify life-threatening conditions in milliseconds and significantly improve work efficiency. This system provides an effective solution to the global shortage of radiologists. The relevant paper was published in the latest issue of JAMA Network Open, a journal under the Journal of the American Medical Association. This AI system has been deployed in 12 hospitals affiliated with Northwestern University in the United States, analyzing nearly 24000 radiology reports over a period of five months in 2024. The data shows that the system has improved the efficiency of generating radiation reports by an average of 15.5%, and some doctors have even increased their efficiency by 40%, with no decrease in accuracy. Subsequent studies have shown that efficiency improvement can reach up to 80% in CT imaging. The saved time allows radiologists to make faster diagnoses, especially in critical cases. Unlike narrow domain AI tools on the market that can only detect a certain type of disease, this AI system can read complete X-ray or CT images and automatically generate personalized reports with a 95% completion rate for doctors to choose from, review, and ultimately confirm. These reports will summarize key findings and provide doctors with auxiliary templates for diagnosis and treatment. In addition to improving efficiency, this AI system can also real-time mark fatal conditions such as pneumothorax (lung collapse), and synchronize cross validation with medical records during the report generation process. Once a critical condition is detected, doctors will be immediately alerted. Researchers say that this is the first time that an AI system has fully demonstrated the advantages of high accuracy and efficiency in various X-ray images covering from head to toe. Unlike relying on large general AI models such as ChatGPT, this AI system is completely self built, and the training data comes entirely from real clinical data within the medical system. This method not only makes AI systems lighter and more accurate, but also significantly reduces dependence on computing resources and runs faster. Research shows that by 2033, the United States is expected to face a shortage of up to 42000 radiologists, with the number of imaging examinations increasing by approximately 5% annually. The new AI system is expected to alleviate this urgent situation and help doctors shorten the delivery time of diagnostic reports from days to hours. Despite its powerful capabilities, the research team emphasizes that its goal is not to replace human doctors. (New Society)