An Oxford AI heart failure detection system developed by researchers at the University of Oxford can identify patients at high risk of heart failure up to five years before the condition develops, according to findings published this week. The system analyzed data from 72,000 patients and achieved 86% accuracy.
The tool works by reading subtle changes in the texture of fat surrounding the heart, a process invisible to doctors reviewing standard scans. When the heart muscle becomes inflamed, the surrounding fat shifts in texture, and the AI detects those patterns from routine CT scans already performed for other medical reasons.
How the Oxford AI Heart Failure Model Works
The system assigns patients to risk categories based on those fat-texture patterns. In the highest-risk group, one in four patients developed heart failure within five years. That rate is 20 times higher than patients the AI classified as low risk, indicating a strong separation between risk groups.
Scale of the Study
The research drew on CT scan data from 72,000 patients, making it one of the larger validation studies for an AI-based cardiac screening tool. Researchers said the model performed consistently across the full dataset, not just in smaller subgroups.
Regulatory Path and NHS Rollout
Oxford is already in discussions with regulators to deploy the tool across National Health Service hospitals in the United Kingdom. The university plans to extend the system to all chest CT scans within months, which would make it a passive screening layer built into existing clinical workflows rather than a separate test.
No additional scans or procedures would be required. Patients who already receive chest CT scans for other conditions would automatically receive a heart failure risk assessment as part of the same imaging session.
Why Early Detection Matters
Heart failure is difficult to treat effectively once significant damage has occurred. Most patients receive a diagnosis only after symptoms appear, at which point the condition has often progressed. A detection window of five years gives clinicians time to intervene with lifestyle changes, medication, or closer monitoring before the heart deteriorates further.
The Oxford team said the approach shifts the clinical model from reaction to prevention. Because the tool runs on CT scans patients are already receiving, it adds no extra burden to patients or imaging departments. Researchers said they expect the system to improve outcomes for a condition that affects millions of people globally, including a growing number across the Middle East.
“Fat around the heart shifts texture when the muscle beneath is inflamed, with the AI reading the patterns invisible to doctors on any current scan.”
University of Oxford Research Team
The development adds to a growing body of work applying artificial intelligence to medical imaging, where pattern recognition at scale has repeatedly outperformed manual review in early detection tasks across cardiology, oncology, and radiology.
