AI-guided heart care increased cholesterol treatment in preliminary randomized trial

Patients evaluated for suspected coronary artery disease were more likely to receive cholesterol-lowering treatment when care used AI-assisted analysis of plaque in heart scans. Cleerly presented the preliminary PARAMOUNT trial results on November 7 at the American Heart Association’s Scientific Sessions in Chicago, reporting that the approach beat stress-test-guided care on its main 180-day measure.

The findings have not been peer reviewed. Cleerly expects full results to be published in early 2027. The main outcome combined diagnostic certainty, cholesterol control and measures of invasive-procedure use; it should not be read as evidence that the strategy prevented heart attacks.

Two routes from chest pain to treatment

PARAMOUNT randomly assigned 530 participants across 21 sites between January 2025 and February 2026. The preliminary analysis used locked data from 523 participants. They were symptomatic outpatients with suspected coronary artery disease, elevated LDL cholesterol of at least 100 mg/dL, and no prior history of the disease.

One group received care guided by AI-based quantitative coronary CT analysis, or AI-QCT: software that measures plaque and assesses its characteristics in CT images. That strategy also included an assessment of whether plaque was associated with reduced blood flow. The comparison group received usual care guided by functional stress testing.

This was a comparison of care strategies, not simply a contest between software and a doctor reading the same scan. The trial was open-label, meaning treatment assignments were not concealed. Women accounted for 300 of the 523 participants in the analysis, or 57%, and the median participant age was 57.

Clearer answers drove the main result

Cleerly reported a win ratio of 2.00 on the prespecified hierarchical composite endpoint, which combines several outcomes in a ranked comparison. The 95% confidence interval was 1.37–2.56. In unmatched patient comparisons, the plaque-based-care patient had the better result 46.7% of the time, versus 23.4% for usual care.

Greater certainty about whether coronary artery disease caused a patient’s chest pain was the largest contributor, with a win difference of 16.6 percentage points. LDL cholesterol reductions also favored the AI-guided strategy. That distinction is important: the headline result reflects clearer diagnoses and changes in care, rather than one standalone measure of disease progression.

A different pattern of invasive procedures

Among patients who underwent invasive coronary angiography, 38% in the AI-guided group did not proceed to a procedure to restore blood flow, compared with 60% in usual care. Those procedures included coronary intervention or bypass surgery. The percentages apply only to patients who underwent angiography, not to everyone assigned to each group.

Trial co-principal investigator Todd Villines said in Cleerly’s announcement that plaque-based care did not increase overall rates of catheterization or procedures to restore blood flow. That separates the reported change in what happened after an invasive test from a claim that more patients were sent for such testing.

In an HCPLive interview, Villines explained the diagnostic goal: stress testing can leave patients wondering whether they have plaque, even when it provides other answers. He also described faster, standardized scan reports as a potential aid to physicians. HCPLive disclosed that Villines consults for Cleerly, HeartFlow and Milestone.