Imagine if you could predict heart failure before it develops.
The researchers at Oxford researchers have developed an AI tool that can predict heart failure at least five years before it clinically develops, using routine cardiac CT scans already performed in the UK's National Health Service (NHS).
Trained and validated on over 70,000 patients across nine NHS Trusts, the system analyses subtle textural changes in fat surrounding the heart—a biological signal invisible to clinicians but strongly associated with early inflammation and future cardiac dysfunction.
Patients flagged as highest risk were 20× more likely to develop heart failure and had a ~25% probability within five years, based on up to a decade of follow‑up data published in the Journal of the American College of Cardiology.
Heart failure is typically diagnosed late, when behaviour change, treatment adherence, and system interventions are hardest and costliest.
The AI tool moves risk identification from crisis to anticipation, enabling earlier clinical action, targeted monitoring, and preventive behaviour support for those who matter most.
Around 350,000 cardiac CT scans are already conducted annually in the UK, and the AI tool requires no new patient behaviour, no new hardware, and minimal workflow friction, critical conditions for large‑scale adoption.
When risk is personalised, quantified, and time‑bounded (five years, not “someday”), clinicians act differently, patients engage earlier, and systems can prioritise resources more intelligently.
This AI doesn’t replace judgement; it reframes it from an abstract aspiration into an operationally actionable moment, years before failure becomes inevitable, with the potential to save lives, and act preventatively.
SOURCE
