Alibaba’s DAMO Academy AI flags acute aortic emergencies on routine CT, published in Nature Medicine

Breakthrough for ER chest pain triage

Alibaba DAMO Academy (阿里巴巴达摩院) and The First Affiliated Hospital, Zhejiang University School of Medicine (浙江大学医学院附属第一医院) have unveiled iAorta, an AI model that identifies acute aortic syndrome (AAS) on standard non-contrast CT scans in seconds. The teams say the system can cut time to confirmed diagnosis to within two hours—crucial for a condition in which each passing hour raises mortality. AAS encompasses aortic dissection, intramural hematoma, and penetrating aortic ulcer—rare but lethal causes of chest pain that are easily missed in chaotic emergency rooms. Why does this matter? Because the gold standard, CT angiography, isn’t always ordered immediately, while non-contrast chest CTs are far more common in first-line triage.

What the study shows

The work has been published in Nature Medicine, a leading peer-reviewed journal—an uncommon feat for hospital-deployed AI from China and a marker of clinical rigor. In real-world ER use, iAorta reportedly scanned routine CTs from more than 10,000 chest-pain patients and precisely flagged 21 true AAS cases, enabling rapid escalation to definitive imaging and surgery when needed. The system integrates into existing radiology workflows to surface high-risk scans to clinicians within seconds, it has been reported. Details on sensitivity, specificity, and external validation cohorts were not disclosed in the announcement, but the publication suggests multi-center evidence and prospective deployment.

Context for China’s digital-health push

The project pairs a top-tier tertiary hospital with Alibaba (阿里巴巴) research, reflecting China’s broader push to embed AI into frontline care amid persistent pressure on emergency services. Domestic deployment can be shaped by geopolitics: U.S. export controls on advanced AI chips have nudged Chinese labs toward optimization on local or indigenous accelerators, potentially influencing scalability and cost. Even so, imaging AI tends to be compute-light at inference, aiding hospital adoption. Wider roll-out will depend on regulatory clearance—typically from China’s National Medical Products Administration—and reproducibility across diverse scanners and populations.

The road ahead

If validated at scale, iAorta could shift chest-pain triage by using the scans clinicians already order to surface a time-critical, often-missed diagnosis. But questions remain: How does performance hold outside high-volume academic centers? Can it reduce door-to-treatment times and mortality in randomized settings? The Nature Medicine publication gives the effort academic weight; now, outcomes-driven trials and regulatory pathways will determine whether this ER assist becomes standard practice in China—and, potentially, abroad.