Can looking directly at coronary plaque help doctors make better decisions for people with chest pain? A US randomised trial has reported encouraging early results, although it has not shown that AI prevents heart attacks or saves lives.
Today’s edition takes an international focus because it is a quieter day for the UK stories covered in this series. That is an editorial choice, not a claim that no British AI research exists. The US PARAMOUNT trial offers a specific healthcare question worth examining without presenting it as a British study or an NHS rollout.
What Was Announced on 7 November?
Cleerly announced primary results from PARAMOUNT at the American Heart Association Scientific Sessions on 7 November 2026. Its release describes a prospective, multicentre, open-label randomised trial involving 530 participants across 21 sites, with preliminary analysis based on 523 participants whose data were locked at that point.
The patients were symptomatic outpatients with suspected coronary artery disease, raised LDL cholesterol and no previous history of coronary artery disease. That is a defined patient group, not a sample of everyone who might have heart problems.
Participants received either care guided by AI-based quantitative coronary CT analysis, including plaque assessment, or usual care guided by conventional functional stress testing. The study compares two care pathways, not simply an AI algorithm against a doctor reading the identical scan.
Clearer Decisions and More Preventive Treatment
According to the company, the trial met its prespecified hierarchical composite endpoint at 180 days, with a win ratio of 2.00 favouring plaque-guided care. Diagnostic certainty was the largest contributor, and LDL-cholesterol lowering also contributed.
A win ratio compares which member of a patient pair does better according to the trial’s ranked outcomes. A value of 2.00 does not mean twice as many lives were saved, half as many heart attacks occurred, or every patient received twice the benefit.
The release also reports increased use of lipid-lowering therapy, relevant to preventive treatment decisions including statin treatment. Use rose from 31% to 75% in the AI-QCT-guided group, compared with 32% to 47% under usual care. Lipid-lowering therapy is a broader category than statins alone, so those figures should not be relabelled as statin-only results.
Why Randomisation Matters, and What It Cannot Settle
Random allocation makes this more informative than an account of a single successful deployment. It helps compare the effect of the two pathways while reducing some differences between the people receiving them.
However, the trial was open-label, meaning the assigned approach was known. Diagnostic certainty, treatment decisions and a short-term composite endpoint are not equivalent to long-term reductions in major clinical events.
A Company Announcement Needs Independent Scrutiny
Cleerly has a commercial interest in the technology. Its own release explicitly says these results have not yet been peer reviewed, and that full publication is expected in early 2027. Independent assessment of the complete methods, endpoint definitions, missing data and longer-term outcomes remains important.
The supplied Business Wire URL returned no readable article text during preparation. The figures above were checked against Cleerly’s dated original release, linked below, rather than inferred from an inaccessible page.
Why Feature It Here?
For UK readers, the useful lesson is about evidence: AI may influence the decisions that follow a scan, and randomised research can test that broader pathway. Whether similar results would apply in NHS settings needs separate consideration.
The good news is an encouraging, testable signal about diagnostic confidence and preventive care. It is worth following while keeping commercial claims, preliminary findings and proven patient outcomes clearly apart.
Sources and Further Reading
- Business Wire: PARAMOUNT trial announcement. Announcement dated 7 November 2026. The supplied page returned no readable text during preparation; the matching original company release was used for verification.
- Cleerly: PARAMOUNT meets primary endpoint at AHA 2026. Published 7 November 2026. Verified company release supplying recruitment, analysis population, treatment figures and the explicit preliminary, not-peer-reviewed caveat.
- ClinicalTrials.gov: PARAMOUNT trial registration. Registration linked by the company announcement. A trial registration is not itself a peer-reviewed report of results.
Less hype. More real-world AI progress.