Family Heart Foundation Study Shows Machine Learning Model Can Help Identify ASCVD Patients at Higher Risk of Elevated Lp(a)
The Family Heart Foundation, a leading research, education and advocacy organization focused on cardiovascular health, has announced the publication of a new peer-reviewed study demonstrating the potential of its FIND Lp(a)® Machine Learning Model to improve identification of patients with atherosclerotic cardiovascular disease (ASCVD) who are more likely to have elevated lipoprotein(a), or Lp(a).
Published in JACC: Advances, the study evaluates the development and initial validation of the FIND Lp(a) Machine Learning Model using data from the Family Heart Database®. The findings indicate that the model can help healthcare systems identify ASCVD patients who have a higher likelihood of elevated Lp(a), potentially enabling more targeted screening while health systems work toward broader implementation of guideline-recommended universal Lp(a) testing.
The model has already been deployed at five large U.S. healthcare systems participating in the Family Heart Foundation’s FIND Lp(a) program. The initiative is designed to use machine learning and electronic medical record data to identify individuals who may be more likely to have high Lp(a), encourage appropriate screening and ultimately connect patients with education and cardiovascular risk-management resources.
The new findings provide clinical evidence supporting the use of predictive analytics as a practical tool for increasing Lp(a) screening in healthcare settings where universal testing has not yet been fully implemented.
Study Demonstrates Screening Enrichment
The peer-reviewed study, titled “FIND Lp(a) MLM: Targeted Screening Enrichment of Elevated Lipoprotein(a) in Atherosclerotic Cardiovascular Disease,” describes the development and initial validation of the machine learning model.
Researchers used information contained in the Family Heart Database to develop a model capable of identifying people with ASCVD who may be more likely to have elevated Lp(a). Patients identified by the model were more than 2.2 times as likely to have high Lp(a), defined in the study as a level of at least 125 nmol/L, compared with the overall population of people with ASCVD represented in the Family Heart Database.
This level of screening enrichment could be important for healthcare organizations seeking to increase Lp(a) testing. Rather than relying exclusively on broad population screening strategies, healthcare systems could use the model to prioritize patients who have a greater likelihood of having elevated Lp(a), allowing screening resources and clinical workflows to be directed toward individuals at potentially increased cardiovascular risk.
The Family Heart Foundation noted that additional validation using retrospective and prospective data from healthcare systems is ongoing. These efforts are expected to provide further information about how the model performs in real-world clinical environments.
Growing Recognition of Lp(a) as a Cardiovascular Risk Factor
Lp(a) is an independent and genetically determined risk factor for cardiovascular disease. Unlike many traditional cardiovascular risk factors, Lp(a) levels are largely determined by genetics and can remain relatively stable throughout a person’s lifetime.
Despite the established association between elevated Lp(a) and cardiovascular risk, awareness and testing remain limited. The Family Heart Foundation estimates that approximately one in five people have high Lp(a), while 99% of people in the United States have never had their Lp(a) level measured.
This gap between the prevalence of elevated Lp(a) and the rate of screening represents a significant challenge for cardiovascular disease prevention.
Patients may undergo conventional cholesterol testing and cardiovascular risk assessments without having their Lp(a) measured. As a result, individuals with elevated Lp(a) may remain unidentified even when they have other indicators of cardiovascular risk or a history of ASCVD.
The Family Heart Foundation believes machine learning could provide healthcare organizations with a way to begin closing this gap by identifying patients who may benefit most from Lp(a) testing.
Supporting the Move Toward Universal Screening
The findings come as recently released U.S. dyslipidemia guidelines recommend Lp(a) screening for all adults. While universal screening could ultimately provide the most comprehensive approach to identifying elevated Lp(a), implementing such recommendations across large healthcare systems can require substantial changes to clinical workflows, electronic health records, provider education and patient engagement.
Diane MacDougall, vice president of Research at the Family Heart Foundation and principal author of the study, emphasized the potential role of targeted screening during this transition.
“Although recently released U.S. dyslipidemia guidelines recommend Lp(a) screening for all adults, integrating this into routine clinical practice will take many years, if not decades,” MacDougall said.
She added that the FIND Lp(a) model could help accelerate adoption of universal screening by identifying people who are most likely to have elevated Lp(a), creating additional opportunities for patients and their healthcare teams to recognize and manage cardiovascular risk.
The approach could therefore serve as a bridge between current levels of Lp(a) testing and the broader goal of universal screening.
FIND Lp(a) Program Combines Data and Clinical Action
The FIND Lp(a) Machine Learning Model is a central component of the Family Heart Foundation’s Flag, Identify, Network, and Deliver™ quality improvement program.
The program is designed around the use of machine learning and electronic medical record data to identify adults who may be at increased likelihood of having elevated Lp(a). Once individuals are identified through the program, participating healthcare systems can offer Lp(a) screening.
Patients whose testing shows high Lp(a) can then receive appropriate clinical care and are invited to engage with the Family Heart Foundation for additional education and support.
This approach is intended to move beyond simply developing a predictive model. The Foundation is integrating the technology into actual healthcare workflows, creating a pathway from patient identification to laboratory testing and, when appropriate, ongoing cardiovascular risk management.
The program is currently being implemented through partnerships with five healthcare systems. Through its Collaborative Learning Network, the Family Heart Foundation and its healthcare partners are prospectively evaluating the real-world performance of the FIND Lp(a) Machine Learning Model.
The organizations are also sharing operational best practices and working to develop sustainable approaches to Lp(a) screening and care.
Real-World Deployment Represents a Key Milestone
The Family Heart Foundation highlighted the real-world deployment of the model as an important achievement in the application of machine learning to clinical practice.
While healthcare organizations are increasingly exploring artificial intelligence, machine learning and predictive analytics, translating these technologies from development environments into routine clinical workflows can be challenging. Barriers may include data quality, integration with electronic health record systems, clinical validation, provider adoption and the development of appropriate processes for acting on predictions.
The FIND Lp(a) project has progressed beyond model development to implementation at multiple large healthcare systems. According to the Foundation, this practical deployment distinguishes the initiative from many machine learning projects that do not progress to sustained clinical use.
By combining predictive analytics with screening and patient engagement, the program is designed to demonstrate how machine learning can support measurable improvements in cardiovascular risk identification.
The ongoing prospective validation work will be particularly important in determining how consistently the model performs across different healthcare environments and patient populations.
Potential Impact on Cardiovascular Risk Management
Identifying patients with elevated Lp(a) can provide clinicians with additional information when assessing cardiovascular risk. Because Lp(a) is genetically determined and may contribute to cardiovascular disease independently of other traditional risk factors, recognizing elevated levels can help inform discussions about overall risk and management.
For patients already diagnosed with ASCVD, identifying high Lp(a) may be particularly relevant because these individuals already have established cardiovascular disease and may benefit from a more comprehensive assessment of factors contributing to their residual risk.
The FIND Lp(a) model is not intended to replace laboratory testing. Instead, its purpose is to help healthcare systems determine which patients should be prioritized for Lp(a) screening within existing clinical workflows.
If validated more broadly, this strategy could allow healthcare organizations to reach patients who might otherwise remain untested while building infrastructure and experience for eventual universal screening.
Building Awareness Around Lp(a)
In addition to its research and technology initiatives, the Family Heart Foundation is working to raise awareness of Lp(a) among clinicians, healthcare administrators, payors and the public.
The organization’s efforts reflect a broader recognition that improving cardiovascular outcomes requires not only new treatments but also better identification of individuals with inherited or underrecognized risk factors.
The high prevalence of elevated Lp(a), combined with the low rate of testing, underscores the scale of the screening challenge. The Foundation believes that data-driven approaches such as FIND Lp(a) could help healthcare systems begin addressing this gap in a practical and scalable way.
The publication in JACC: Advances provides peer-reviewed support for the model’s development and initial validation while ongoing studies continue to assess its performance in real-world healthcare settings.
As the five participating healthcare systems continue prospective validation through the Collaborative Learning Network, the project may provide additional insights into how predictive analytics can be integrated into cardiovascular prevention programs.
The Family Heart Foundation’s new study represents a step forward in the use of machine learning to support Lp(a) screening among people with ASCVD. By identifying patients who are more than twice as likely to have high Lp(a) compared with the broader ASCVD population in the Family Heart Database, the FIND Lp(a) Machine Learning Model demonstrates the potential value of targeted screening enrichment.
The initiative also illustrates how machine learning can be connected to clinical action rather than functioning solely as a research or analytics tool. Through the FIND Lp(a) program, healthcare systems can use model-generated insights to identify patients, offer screening and provide appropriate education and support to those with elevated Lp(a).
With universal Lp(a) screening increasingly emphasized in cardiovascular guidelines, tools capable of helping health systems implement screening strategies could play an important role during the transition toward broader testing.
Further retrospective and prospective validation will be needed to establish the model’s performance across diverse healthcare settings. Nevertheless, the publication and real-world deployment of FIND Lp(a) represent meaningful progress in applying predictive analytics to an important but frequently overlooked cardiovascular risk factor.
The Family Heart Foundation’s work ultimately aims to increase awareness, expand Lp(a) screening and help ensure that more people with elevated genetically determined cardiovascular risk are identified and connected with appropriate care.
About the Family Heart Foundation
The Family Heart Foundation® is a nonprofit research and advocacy organization that receives contributions and sponsorships from individuals, foundations, and pharmaceutical companies. The FIND Lp(a) Initiative is supported by Novartis, although Novartis plays no role in program conduct. The Family Heart Foundation is a pioneer in the application of real-world evidence, patient-driven advocacy, and multi-stakeholder education to help prevent heart attacks and strokes caused by familial hypercholesterolemia (FH) and elevated lipoprotein(a), or Lp(a), two common genetic disorders that have an impact across generations.
The Family Heart Foundation conducts innovative research to break down barriers to diagnosis and management of inherited lipid disorders; educates patients, providers, and policy makers; advocates for change; and provides hope and support for families impacted by heart disease and stroke caused by FH, HoFH, and elevated Lp(a). The organization was founded in 2011 as the FH Foundation.