Elucid Launches Landmark AI-PREDICT Study to Advance Lesion-Level Cardiovascular Risk Assessment
Artificial intelligence is rapidly transforming cardiovascular medicine by enabling clinicians to move beyond traditional methods of assessing heart disease risk. One of the latest initiatives in this evolving field comes from Elucid, a company specializing in AI-powered coronary CT angiography (CCTA) analysis, which has announced the enrollment of the first patient in AI-PREDICT, a landmark international clinical study designed to redefine how cardiovascular risk is evaluated.
The multicenter retrospective study represents an important milestone in precision cardiology, aiming to establish a new lesion-centric approach to identifying coronary artery disease (CAD) risk. Rather than assessing a patient’s overall cardiovascular profile alone, AI-PREDICT seeks to determine which individual coronary plaque is most likely to trigger a future heart attack. By combining advanced artificial intelligence with quantitative imaging analysis, the study hopes to provide physicians with more precise information that could improve prevention strategies and personalize treatment decisions.
Moving Beyond Traditional Cardiovascular Risk Assessment
Heart disease continues to be the leading cause of death worldwide despite decades of medical advances, improved medications, and significant investments in cardiovascular research. Although clinicians have become increasingly effective at managing common risk factors such as hypertension, diabetes, elevated cholesterol, and smoking, accurately predicting which patients will experience a heart attack—and more importantly, which coronary plaque will cause it—remains a major clinical challenge.
Traditional cardiovascular risk assessments primarily evaluate patient-level characteristics, including age, family history, blood pressure, cholesterol levels, diabetes status, lifestyle factors, and the overall extent of coronary artery narrowing. While these methods provide valuable information, they often fail to identify the specific plaque that may eventually rupture and cause an acute myocardial infarction (MI).
Research has shown that many heart attacks originate from plaques that appear only moderately narrowed during routine imaging. These plaques may not produce noticeable symptoms or significantly restrict blood flow before suddenly becoming unstable and triggering a cardiac event.
The AI-PREDICT study is designed to address this limitation by focusing on lesion-level analysis rather than relying solely on patient-level risk assessment.
First Patient Enrollment Marks Major Milestone
Elucid announced that the first patient has now been enrolled in AI-PREDICT, officially launching what is expected to become one of the largest international studies dedicated to lesion-specific cardiovascular risk stratification.
Enrollment has already begun across three leading academic medical centers among more than twenty planned research sites located throughout North America, Europe, and Asia.
The first participating institutions include:
- Emory University
- Medical University of South Carolina (MUSC)
- Centro Cardiologico Monzino in Milan, Italy
Each institution is recognized internationally for excellence in cardiovascular imaging, radiology, and cardiac research.
The study is expected to expand significantly as additional hospitals and research centers begin recruiting eligible participants over the coming months.
A Large International Research Collaboration
AI-PREDICT has been designed as a retrospective, multicenter international clinical investigation that will ultimately include approximately 1,000 participants.
The study population will consist of two carefully selected groups:
The first group will include patients who experienced a myocardial infarction within 36 months following their baseline coronary CT angiography examination.
The second group will include clinically matched control patients who remained free from cardiovascular events during the same observation period.
By comparing these two populations, investigators hope to identify the unique characteristics that distinguish dangerous plaques from those that remain stable over time.
This comparison could help establish entirely new biomarkers capable of predicting future cardiovascular events with greater accuracy than currently available approaches.
Building a Comprehensive Lesion-Level Database
One of the most ambitious aspects of AI-PREDICT is its focus on creating an extensive database of individual coronary artery lesions rather than simply collecting patient-level information.
Every coronary plaque identified within enrolled participants will undergo detailed evaluation using advanced imaging software developed by Elucid.
All coronary CT angiography scans will be analyzed through a centralized imaging core laboratory operating independently from participating clinical sites.
Importantly, analysts within the core laboratory will remain blinded to patient outcomes throughout the evaluation process to minimize bias and strengthen scientific validity.
The analysis will utilize Elucid’s proprietary technologies, including:
- Plaque-IQ™
- FFR-CT software
Together, these technologies evaluate numerous imaging characteristics that may influence cardiovascular risk.
Researchers will examine:
- Plaque morphology
- Plaque composition
- Anatomical structure
- Physiological blood flow characteristics
- Coronary artery narrowing
- Functional significance of lesions
Artificial intelligence algorithms will assist investigators in extracting detailed quantitative measurements that would be difficult or impossible to perform manually on such a large scale.
Understanding Why Some Plaques Become Dangerous
One of the central scientific questions driving AI-PREDICT is deceptively simple yet enormously important.
Among the many plaques that may exist inside a patient’s coronary arteries, why does only one eventually rupture and cause a heart attack while neighboring plaques remain harmless?
Answering this question has challenged cardiovascular researchers for decades.
According to investigators participating in the study, lesion-by-lesion analysis may finally provide the detailed information necessary to distinguish unstable plaques from stable ones.
Dr. Gianluca Pontone, principal investigator for the Milan study site, explained that the research attracted early participation because of its unique focus on understanding the characteristics of the individual plaque responsible for causing clinical events.
Rather than simply evaluating overall disease burden, AI-PREDICT seeks to isolate the precise biological and structural features that separate culprit lesions from non-threatening plaques within the same patient.
Why Current Diagnostic Methods Have Limitations
Current cardiovascular imaging technologies have greatly improved physicians’ ability to diagnose coronary artery disease.
Coronary CT angiography has become one of the most valuable non-invasive tools for visualizing coronary anatomy and detecting arterial narrowing.
However, conventional interpretation often emphasizes the degree of stenosis—or narrowing—present within coronary arteries.
Studies have demonstrated that plaque vulnerability depends on much more than narrowing alone.
Several characteristics contribute to whether a plaque may become unstable, including:
- Lipid-rich composition
- Inflammatory activity
- Fibrous cap thickness
- Plaque remodeling
- Calcification patterns
- Local blood flow dynamics
Many dangerous plaques do not produce severe stenosis before rupture.
Consequently, relying primarily on narrowing measurements may overlook lesions that eventually become responsible for heart attacks.
Artificial intelligence provides an opportunity to evaluate numerous imaging features simultaneously, potentially improving clinicians’ ability to recognize high-risk plaques earlier.
Leadership by International Cardiovascular Experts
AI-PREDICT brings together many of the world’s leading authorities in cardiovascular imaging, plaque biology, radiology, and artificial intelligence.
The study is led by Dr. Jagat Narula, President of the World Heart Federation and an internationally recognized expert in cardiovascular imaging and plaque research.
Serving as co-principal investigator is Dr. Carlo N. De Cecco, Professor of Radiology and Biomedical Informatics and Director of the Cardiothoracic Imaging Division at Emory University School of Medicine.
Dr. De Cecco has contributed extensively to the development of quantitative cardiac CT imaging and the integration of artificial intelligence into cardiovascular diagnostics.
The study’s Steering Committee is jointly chaired by Dr. Amir Ahmadi, Clinical Associate Professor of Medicine (Cardiology) at the Icahn School of Medicine at Mount Sinai and Lead Scientific Advisor at Elucid, alongside Dr. Michael Hadley, System Director of Advanced Cardiac Imaging at Northwell Health.
Together, the leadership team combines expertise spanning cardiology, radiology, computational imaging, clinical research, and AI-driven medical technologies.
Shifting Toward Personalized Cardiovascular Care
According to Dr. Jagat Narula, AI-PREDICT represents a significant conceptual shift in cardiovascular medicine.
Historically, physicians have estimated cardiovascular risk at the patient level using broad statistical models based on population studies.
Yet heart attacks occur because individual plaques rupture—not because patients belong to a particular statistical category.
AI-PREDICT seeks to bridge this disconnect by focusing attention directly on lesion-level biology and imaging characteristics.
If successful, the study could support a more personalized approach to coronary artery disease prevention in which treatment decisions are guided by the specific behavior of individual plaques rather than generalized risk estimates.
Comparing Culprit and Stable Lesions
Another distinctive feature of AI-PREDICT is its research design.
Investigators will compare:
- Culprit lesions responsible for myocardial infarction
- Non-culprit plaques within the same patient
- Stable lesions in matched control patients
This three-way comparison offers a unique opportunity to understand what differentiates plaques that become clinically dangerous from those that remain stable.
According to Dr. Amir Ahmadi, the study may provide the first rigorous large-scale framework for evaluating lesion-specific cardiovascular risk across diverse patient populations.
Such insights could eventually support more individualized prevention strategies for patients living with coronary artery disease.
Supporting the Vision of CCTA 3.0
The AI-PREDICT study aligns closely with Elucid’s broader vision of advancing coronary CT angiography into what the company describes as “CCTA 3.0.”
Traditional coronary CT primarily focused on identifying whether coronary arteries contained blockages.
More recent advances have expanded imaging capabilities to evaluate plaque composition and coronary physiology.
Elucid envisions the next generation of CCTA combining advanced artificial intelligence, quantitative plaque analysis, physiological modeling, and predictive analytics to provide truly personalized cardiovascular care.
Under this framework, clinicians could potentially identify the specific lesions most likely to become unstable and intervene before patients experience heart attacks.
Potential Impact on Future Clinical Practice
Should AI-PREDICT successfully identify reliable lesion-level biomarkers, the findings could influence multiple areas of cardiovascular medicine.
Potential future applications include:
- Improved individualized cardiovascular risk prediction
- Earlier identification of vulnerable coronary plaques
- Enhanced decision-making for preventive therapies
- Better selection of patients requiring intensive monitoring
- More personalized cholesterol-lowering treatment strategies
- Refined use of advanced cardiac imaging
- Improved allocation of healthcare resources
The study may also accelerate broader adoption of artificial intelligence in cardiovascular diagnostics by demonstrating how machine learning can extract clinically meaningful information from routine coronary CT scans.
The enrollment of the first patient in AI-PREDICT marks the beginning of an ambitious international effort to transform cardiovascular risk assessment through artificial intelligence and advanced imaging. By shifting the focus from generalized patient risk to detailed lesion-level analysis, Elucid and its global network of research collaborators hope to answer one of cardiology’s most important unanswered questions: which coronary plaque is most likely to cause the next heart attack.
As approximately 1,000 patients are enrolled across more than 20 research sites spanning the United States, Europe, and Asia, the study is expected to generate one of the most comprehensive datasets ever assembled on coronary plaque characteristics. Combined with AI-powered analysis using Plaque-IQ™ and FFR-CT technologies, the findings may lay the foundation for a new era of precision cardiovascular care in which clinicians can identify vulnerable lesions earlier, tailor preventive treatments more effectively, and ultimately reduce the global burden of coronary artery disease.
