1 September 2026
AI reads heart disease signs from routine ECG in under two seconds
First reported
The Guardian ran this on , a day before the next source picked it up.
- Imperial College London researchers created an AI model that analyzes routine heart electrical recordings and identifies heart failure in 81% of cases and valve disease in 90%.
- The tool extracts information from standard ECGs that human doctors cannot typically see, potentially allowing high-risk patients to skip months of waiting for follow-up ultrasound scans.
- The AI cannot diagnose on its own but flags patients for urgent further testing. Trials are starting at six hospitals with plans to integrate it into UK NHS hospitals within two years.
- ECGs are among the most common medical tests worldwide, performed about one billion times annually, making this application potentially relevant to millions of patients.
How it was covered
The Rundown AIRowan Cheung
Imperial College London researchers developed an AI model that reads routine ECGs in under two seconds and spots heart failure and valve disease undetectable by doctors, flagging heart failure in 81% of cases and valve disease in 90%. The team is starting trials at six hospitals with a goal of incorporating the AI into routine NHS use within two years.
Reported by The Guardian