Heartflow DECIDE registry shows 34% fewer CVD events with AI plaque tool
Heartflow has presented one-year outcomes from its DECIDE registry at the American Heart Association Scientific Sessions 2026 in San Francisco, reporting that patients assessed with its AI-powered Heartflow Plaque Analysis experienced 34% fewer major cardiovascular events and half the rate of myocardial infarction compared with those assessed by coronary computed tomography angiography alone.
The registry enrolled more than 6,000 well-matched patients and tracked a composite endpoint of cardiovascular death, MI, revascularisation, and stroke over twelve months. By quantifying total plaque volume and staging disease severity, the platform enabled clinicians to adjust drug regimens earlier: more than half of patients had their management changed within 90 days, and approximately 50% reached a target LDL-C of below 70 mg/dL by that point, compared with around 35% in the CCTA-only cohort. That lipid control was sustained through one year, with an average LDL-C reduction of 18.7 mg/dL in patients whose management was altered.
Leslee Shaw, Director of the Blavatnik Family Research Institute at the Icahn School of Medicine at Mount Sinai and co-lead investigator of the DECIDE registry, said the findings directly link non-invasive plaque imaging to meaningful risk reduction. "By directly linking this non-invasive approach to targeted lipid lowering, it helps bridge diagnostic imaging with meaningful risk reduction in everyday practice," she said.
GAMEFILM: hidden risk in retired professional athletes
Heartflow also presented findings from its GAMEFILM registry, which evaluated coronary artery disease in 200 retired NFL players using CCTA paired with Plaque Analysis. Despite fewer than 10% reporting exertional symptoms, detectable coronary plaque was present in 85% of participants, while only 13% had a prior CAD diagnosis. Nearly 90% of detected plaque was soft and non-calcified, a composition associated with greater vulnerability than calcified plaque and one that conventional calcium scoring cannot identify. Some 42% had obstructive CAD, and retired linemen exhibited plaque volumes at the 63rd age-adjusted percentile on average, compared with the 32nd percentile for non-linemen.
Notably, 88% of players not currently on lipid-lowering therapy said they would be willing to start treatment after reviewing their AI plaque results with a physician, a finding Heartflow positions as evidence that personalised imaging data can drive behaviour change. The GAMEFILM registry is expanding beyond the NFL to include former NBA, NHL, and MLB players, as well as professional staff and US college sports athletes.
Market context and regulatory standing
Heartflow Plaque Analysis holds FDA clearance and the company reports 95% agreement with intravascular ultrasound in a prospective international trial, a claim that, if borne out at scale, distinguishes it from software tools validated only retrospectively. The Plaque Staging framework is said to be validated across 23,000 patients with up to 16 years of follow-up, offering a longer evidence horizon than most competing coronary imaging analytics platforms.
The broader market for AI-assisted cardiac imaging is expanding rapidly, with a number of companies developing automated plaque characterisation and stenosis-assessment tools built on coronary CTA. Heartflow's competitive argument centres on its proprietary dataset of more than 200 million annotated CTA images and its FFR-CT workflow, which is backed by ACC/AHA guidelines and adopted by more than 1,800 institutions globally. The DECIDE prospective outcomes data, once peer-reviewed and published, could strengthen reimbursement arguments in markets where payers still require clinical-outcome evidence rather than diagnostic accuracy alone. Clinicians and investors will look closely at the full dataset, including subgroup breakdowns and adverse event rates, ahead of any formal publication.