Beyond the Scan: How AI is Redefining Preventative Cardiology
The persistent challenge in cardiology isn’t necessarily treating heart disease, but finding it before it causes irreversible damage. While public health campaigns emphasize lifestyle factors, and clinical guidelines detail screening for known high-risk individuals, a significant portion of cardiovascular disease develops silently, undetected until a crisis occurs. McLaren Health Care’s recent launch of an artificial intelligence-powered screening tool, in partnership with Bunkerhill Health, isn’t simply another technological advancement; it represents a shift in how we conceptualize preventative care, moving from reactive diagnosis to proactive identification of risk within existing healthcare encounters. The system, now operational across all 12 McLaren hospitals in Michigan, analyzes routine CT scans – those ordered for reasons other than heart concerns – to flag patients potentially harboring undiagnosed cardiovascular disease.
This piece references the The Detroit News report.
The core innovation lies in repurposing data already collected. Currently, a coronary calcium score, a key indicator of heart disease risk, is typically ordered by a physician actively investigating a patient’s cardiovascular health. What McLaren and Bunkerhill Health have achieved is the ability to calculate this score, and assess aortic valve calcification, from scans already performed for unrelated issues like trauma or pneumonia. This isn’t about creating new tests, but extracting valuable information from tests that are already standard practice. As Dr. Samar Kazziha, McLaren Health Care chief medical director of the McLaren Heart and Vascular Institute, stated during Wednesday’s press conference, “This is a very important way of finding the disease before it causes any problems for the patient long-term or short-term.” The system doesn’t replace a cardiologist’s assessment, but acts as an early warning system, alerting both primary care physicians and patients to potential concerns.
The projected impact is substantial. McLaren estimates that 8-10% of patients undergoing CT scans will be flagged as having previously unknown cardiovascular risk factors. This translates to at least 3,000 patients annually within the McLaren system alone. Dr. Justin Klamerus, McLaren Health Care executive vice president and chief clinical officer, emphasized the goal: “to be partners with our patients to make sure that risk factors are mitigated, that interventions are given to help control the things that help prevent patients from ultimately dying from heart disease or suffering from the morbidity associated with heart disease, stroke.” This focus on “orphaned” patients – those without a consistent cardiologist or primary care physician actively monitoring their heart health – is particularly noteworthy. The system aims to bridge a critical gap in care, identifying individuals who might otherwise remain unaware of their risk until a serious event occurs.
However, it’s crucial to understand what this AI doesn’t do. Headlines proclaiming a revolution in heart disease detection often oversimplify the process. The AI doesn’t provide a diagnosis; it provides a risk assessment. A flagged scan triggers an alert, prompting a conversation between the patient and their physician, potentially leading to further testing and intervention. Furthermore, the accuracy of any AI model is dependent on the data it’s trained on. While Bunkerhill Health has partnered with over a dozen health systems, including the Cleveland Clinic, the generalizability of the model across diverse populations remains a key consideration.
Limitations to Consider
The 8-10% figure, while significant, is an expectation based on initial data. The actual percentage could vary depending on the patient population and the specific CT scanning protocols used at each McLaren hospital. There’s also the potential for false positives – scans flagged as high-risk when, in reality, the patient doesn’t have significant cardiovascular disease. This could lead to unnecessary anxiety and further testing. Equally important is the question of follow-through. An alert is only effective if it’s acted upon. Ensuring that primary care physicians are equipped to interpret the AI’s findings and initiate appropriate interventions will be critical to the program’s success. The system also relies on patients being engaged in their own care and willing to follow up on recommendations.
The broader implications extend beyond McLaren Health Care. This initiative highlights a growing trend: the integration of AI into routine clinical workflows to enhance preventative care. The success of this program in Michigan will likely influence other health systems across the country to explore similar applications of AI-powered screening. The question now isn’t if AI will play a larger role in preventative cardiology, but how it will be implemented responsibly and equitably. Will this technology exacerbate existing health disparities, or will it help to close the gap in access to preventative care?
The next crucial step is longitudinal data collection. McLaren needs to track the outcomes of patients flagged by the AI – their adherence to recommended interventions, their rates of cardiovascular events, and their overall health trajectories. This data will be essential to refine the AI model, validate its effectiveness, and demonstrate its value to both patients and healthcare providers. Specifically, researchers should investigate whether early detection through this AI-powered system translates into a measurable reduction in heart attacks, strokes, and cardiovascular mortality. And, importantly, they need to monitor for any unintended consequences, such as increased healthcare costs or disparities in access to follow-up care. Will we see a demonstrable shift in the curve of cardiovascular disease incidence in Michigan, and if so, can that success be replicated elsewhere?
