Superhuman AI tool spots heart disease in under two seconds

[BringItOnRhodes43]

Researchers have developed an AI tool that can analyze a routine ECG and flag signs of heart disease in under two seconds, extracting far more information from the reading than a human doctor typically can spot by eye. The tool was trained on data from millions of patients and is specifically aimed at identifying people most at risk of heart failure and heart valve disease so they can be prioritized for further scans faster. Professor Fu Siong Ng of Imperial College London, involved in the research, said patients can currently wait months for a heart ultrasound after being referred, and this tool could help fast track the ones most likely to actually need one

Importantly, the tool can't diagnose heart failure or valve disease on its own, it just flags a strong likelihood that then gets confirmed through a proper echocardiogram. That distinction matters a lot practically, since it's meant to reorder an existing queue rather than replace the actual diagnostic process. Separately, researchers at the University of Tokyo and the Institute of Science Tokyo are also exploring whether AI analysis of short facial videos could help flag related risk factors like high blood pressure and undiagnosed type 2 diabetes

Millions of people currently walk around with high blood pressure or diabetes without knowing it, so any tool that helps surface hidden risk earlier could genuinely matter at scale. Curious what people think about AI tools like this being used specifically to triage and prioritize within existing healthcare queues rather than replace clinical diagnosis outright

'Superhuman' AI tool spots heart disease in less than 2 secondswww.theguardian.com

[UltraWarren94]

Facial video analysis for spotting high blood pressure sounds almost too good to be true, but if it actually pans out that's a genuinely low friction way to catch risk factors people wouldn't otherwise think to get checked. Curious how accurate that specific approach actually turns out to be at scale. Passive screening through something as simple as a video has real appeal

[SchrodingersCat]

Millions of undiagnosed cases of high blood pressure and diabetes is a genuinely staggering number when you actually sit with it for a second.

Early detection tools like this matter less for the technology itself and more for just getting people into a doctor's office sooner. Feels like the actual bottleneck has always been awareness rather than diagnostic accuracy

[Hollow]

Extracting that much diagnostic information from a routine ECG in just two seconds is a genuinely impressive speed improvement over standard human review. Would be curious to see the actual false positive rate though, since triage tools that flag too many people defeat their own purpose. Speed matters less than precision here