AI algorithm flags heart disease directly from standard ECG scans

Medical researchers have created an artificial intelligence system that can analyze ordinary electrocardiogram scans in roughly two seconds to spot hidden signs of serious heart conditions. While traditional electrical heart readings usually cannot detect structural issues like heart failure or damaged valves, this software extracts subtle patterns invisible to human eyes.

In clinical testing with tens of thousands of individuals, the technology correctly flagged up to 90 percent of structural valve issues and most heart failure cases. The tool was trained on millions of historical tests. Doctors could run the algorithm on routine scans taken for unrelated reasons, potentially catching undiagnosed heart problems before symptoms worsen.

Normally, confirming these cardiac issues requires a specialized ultrasound scan, which often involves long waiting lists. As Dr Sonya Babu-Narayan from the British Heart Foundation told The Guardian, while the software will not spot every condition, it can fast-track high-risk individuals for urgent care. Accelerating this screening process allows vulnerable patients to begin life-saving medications months earlier than usual.

However, the tool is not a standalone diagnostic replacement and still requires traditional ultrasound follow-ups to verify its predictions. It remains uncertain how healthcare systems will handle the potential influx of priority referrals, or whether false alarms could unnecessarily strain hospital resources.