Recently, the First Affiliated Hospital of Zhejiang University School of Medicine and Alibaba Damo Hospital jointly released the AI model iAorta for chest pain emergency scenarios. This model can identify whether there is a risk of acute aortic syndrome in the cardiovascular system on conventional plain CT within a few seconds. The research results were recently published in the international journal Nature Medicine. iAorta has been deployed in the first 10 hospitals in Zhejiang and will be further promoted nationwide in the future. The so-called acute aortic syndrome, which appears "invisible" under plain CT scan, refers to the destruction of the integrity of the aortic wall. The most common diseases are aortic dissection, aortic wall hematoma, and aortic penetrating ulcer. Among them, the mortality rate of aortic dissection is very high, and the 24-hour mortality rate of patients in the acute phase can reach 20% -30%. This requires doctors to diagnose early, diagnose quickly, and treat in a timely manner to save such patients. ”Zhang Hongkun, Director of the Department of Vascular Surgery at the First Affiliated Hospital of Zhejiang University School of Medicine, said. Under normal circumstances, experienced doctors can easily diagnose typical patients with acute aortic syndrome. However, the symptoms of most patients are not typical, manifested as abdominal pain, back pain, and other symptoms, resulting in a misdiagnosis rate of up to 40%. Currently, there is no biological indicator that can be used to diagnose acute aortic syndrome. However, its identification is not difficult as long as an enhanced CT scan of the aorta, namely angiography, is performed to diagnose it. However, in emergency situations, due to various factors such as atypical symptom manifestations, certain side effects of contrast agents, and the fact that most hospitals stop performing enhanced CT late at night, doctors mostly ask patients to undergo plain CT scans, and there are very few cases of enhanced CT scans. However, conventional plain CT scan is not an effective means of detecting acute aortic syndrome. Can we use existing commonly used tests to screen for acute aortic syndrome as soon as possible? ”With this kind of thinking in mind, the First Affiliated Hospital of Zhejiang University School of Medicine, in collaboration with Alibaba Damo Institute, has been developing the AI model iAorta since 2022. Xu Minfeng, a senior algorithm expert at Alibaba Damo Institute, introduced that the team has developed the AI model iAorta based on their technical accumulation in the direction of "flat scan CT+AI". Embed iAorta into the hospital's PACs system, and when the plain scan CT is uploaded to the PACs system, doctors start reading the images while iAorta synchronously detects them. It can detect structural changes in the blood vessel wall within seconds, accurately identifying acute aortic syndrome. Once highly suspected of acute aortic syndrome, it will issue a warning and mark which part of the body is at risk, facilitating doctors to make targeted judgments on whether to confirm the condition through enhanced CT. The research on early diagnosis and treatment of aortic diseases has been ongoing for over three years. During this period, the team selected 20000 cases from 8 hospitals in China, including patients who had undergone plain CT scans and patients who had undergone enhanced CT scans, and then inputted the data into iAorta for validation. The verification results show that iAorta has a sensitivity of 97% and a specificity of 94% for recognition. The team found that iAorta's sensitivity and specificity surpassed those of senior doctors by comparing its viewing data with over 2000 cases of viewing data from 4 junior, 4 mid senior, and 3 senior doctors. Meanwhile, with the assistance of iAorta, junior doctors can increase their sensitivity to film reading by more than 40%. In addition, the team conducted a retrospective study on 130000 patients who visited the First Affiliated Hospital of Zhejiang University School of Medicine, Shaoxing Central Hospital, and Quzhou People's Hospital in real-life scenarios. The review results show that iAorta has a sensitivity of 92.6% and a specificity of 99.2%. Among these 130000 patients, iAorta diagnosed a total of 248 patients with acute aortic syndrome. If there is no diagnosis from iAorta, the missed diagnosis rate can reach 48.8%. ”Zhang Hongkun said that after iAorta intervention, the missed diagnosis rate was only 4.8%. He also introduced that some of the 130000 patients who were previously missed were diagnosed with acute aortic syndrome because the initial plain CT scan did not identify the true symptoms. After changes in the condition and various examinations, the diagnosis was finally made, with an average diagnosis time of 11.4 hours. After iAorta intervention, the diagnosis time was shortened to 1.1 hours. Subsequently, the team involved iAorta in a clinical observational study. Among the 13000 patients tested, 11 were diagnosed with acute aortic syndrome. It is worth mentioning that without iAorta intervention, doctors only identified 2 cases of acute aortic syndrome patients through individual film reviews. Through iAorta assistance, doctors identified a total of 9 patients with acute aortic syndrome. Zhang Hongkun emphasized that in a real diagnosis and treatment environment, all imaging films including abdominal plain CT, chest plain CT, lumbar CT, etc., as long as they involve the aortic region, iAorta can make corresponding identification. Finally, the team conducted a prospective intervention study by deploying iAorta in hospitals. This study began in December 2024 and has tested over 15000 patients. A total of 22 patients with acute aortic syndrome were identified, with a specificity of 99.4% and a sensitivity of 95.5%. Among them, 21 patients achieved successful AI warning. At present, the first batch of 10 hospitals in Zhejiang have taken the lead in deploying the iAorta model and formed a fast track for diagnosis, transportation, and treatment. The research team is further lowering the technology access threshold and accelerating the promotion of iAorta to more regions across the country. (New Society)