HONG KONG MEDICAL RESEARCHERS UNVEIL AI BLOOD TEST CAPABLE OF PREDICTING SERIOUS CARDIOVASCULAR PROBLEMS UP TO 15 YEARS IN ADVANCE

A groundbreaking artificial intelligence tool developed by researchers at the LKS Faculty of Medicine of the University of Hong Kong (HKUMed) promises to revolutionize cardiovascular disease prevention, offering the potential to identify serious health risks up to 15 years before clinical symptoms manifest. This innovative system, named CardiOmicScore, leverages a single blood test to analyze a complex array of biological markers, providing a deeply personalized and predictive assessment of an individual’s future likelihood of developing six major cardiovascular conditions. The findings, published in the prestigious journal Nature Communications, herald a significant leap forward in shifting the paradigm of healthcare from reactive treatment to proactive, early intervention.

The six cardiovascular diseases (CVDs) for which CardiOmicScore can estimate future risk include coronary artery disease, stroke, heart failure, atrial fibrillation, peripheral artery disease, and venous thromboembolism. This comprehensive predictive capability addresses a critical unmet need, as cardiovascular diseases remain the leading cause of death globally, claiming an estimated 19.8 million lives in 2022 alone, according to the World Health Organization. The sheer scale of this public health crisis underscores the urgency for advanced diagnostic and preventive strategies.

The Limitations of Traditional Risk Assessment

Current clinical practices for assessing cardiovascular risk typically rely on a combination of well-established factors such as age, blood pressure, smoking history, cholesterol levels, and Body Mass Index (BMI). While these indicators are invaluable for providing a general overview of an individual’s health status and guiding initial management strategies, they often fail to capture the subtle, early biological changes that occur deep within the body long before any outward symptoms become apparent. This diagnostic lag can mean that individuals are not identified as being at high risk until the most opportune window for effective prevention has already begun to narrow, diminishing the potential impact of interventions.

Another avenue for risk assessment, genetic testing, offers insights into an individual’s inherited predisposition to certain diseases. Polygenic risk scores, for instance, aggregate the influence of numerous genetic variants to provide a measure of genetic susceptibility. However, a person’s genetic makeup is largely immutable, established at birth. Consequently, genetic risk scores cannot fully account for the dynamic and ongoing influences that shape health over a lifetime. These influences include significant factors such as dietary habits, physical activity levels, the natural process of aging, the presence of other illnesses, and exposure to environmental factors, all of which can profoundly impact an individual’s current biological state and future disease risk.

CardiOmicScore: A Multi-Layered Biological Snapshot

CardiOmicScore was meticulously designed to overcome these limitations by providing a more dynamic and current reflection of an individual’s internal biological environment. The HKUMed team employed sophisticated deep learning techniques to integrate multiple layers of biological information, a methodology known as multiomics. This approach transcends the confines of single-discipline analysis by drawing data from diverse biological domains, including genomics, proteomics, and metabolomics.

Genomics, the study of an organism’s complete set of genes, provides the foundational understanding of an individual’s inherited blueprint. Proteomics, on the other hand, focuses on the proteome – the complete set of proteins produced or modified by an organism or system. Proteins are the workhorses of the cell, carrying out a vast array of essential functions, including metabolic processes, cellular signaling, and immune responses. Metabolomics delves into the realm of metabolites, which are small molecules produced or utilized during metabolic processes. These metabolites, such as glucose, amino acids, and lipids, serve as critical indicators of an organism’s physiological state, reflecting its response to diet, disease, and environmental exposures.

The researchers harnessed the power of multiomics by analyzing a substantial dataset from the UK Biobank, a large-scale biomedical database containing in-depth genetic and health information from half a million participants. Their AI model was trained on data measuring 2,920 circulating proteins and 168 metabolites extracted from blood samples. By integrating these diverse molecular signatures, CardiOmicScore can construct a remarkably detailed and nuanced snapshot of a person’s current biological state. This detailed molecular profile is capable of revealing subtle shifts in immune system activity, metabolic pathways, and vascular health that may precede the onset of any noticeable symptoms, offering a crucial early warning system.

Professor Zhang Qingpeng, an Associate Professor in the Department of Pharmacology and Pharmacy at HKUMed and a lead researcher on the project, articulated the significance of this multi-layered approach. "Genes determine where we start – they define our baseline health risk," Professor Zhang explained. "However, proteins and metabolites reflect our current physical health. Our AI tool is designed to decode these complex molecular signals, enabling doctors and patients to identify risks much earlier, which can potentially change the trajectory of disease through timely lifestyle modifications and early prevention." This statement underscores the pivotal role of CardiOmicScore in bridging the gap between genetic predisposition and actionable preventive measures.

Predicting a Spectrum of Cardiovascular Diseases

The rigorous validation of CardiOmicScore demonstrated its remarkable ability to translate intricate molecular measurements into personalized risk estimates for a range of cardiovascular diseases. The system’s performance significantly surpassed that of conventional polygenic risk scores, highlighting the added value of integrating proteomic and metabolomic data. Furthermore, the accuracy of the model saw further improvement when researchers incorporated readily available clinical information such as age and gender, creating a more robust and comprehensive predictive tool.

The specific cardiovascular conditions that CardiOmicScore is designed to predict represent some of the most significant global health challenges:

• Coronary Artery Disease (CAD): This condition occurs when the heart’s arteries narrow or become blocked, restricting blood flow to the heart muscle and increasing the risk of heart attack.
• Stroke: A stroke happens when the blood supply to the brain is interrupted, either by a blockage (ischemic stroke) or a rupture of a blood vessel (hemorrhagic stroke), leading to brain damage.
• Heart Failure: This chronic condition occurs when the heart muscle doesn’t pump blood as well as it should, leading to symptoms like fatigue, shortness of breath, and swelling in the legs and feet.
• Atrial Fibrillation (AFib): A type of irregular heartbeat that can significantly increase the risk of stroke, heart failure, and other heart-related complications.
• Peripheral Artery Disease (PAD): This condition involves the narrowing of arteries that supply blood to the limbs, most commonly the legs, leading to pain, cramping, and reduced circulation.
• Venous Thromboembolism (VTE): This encompasses two serious conditions: deep vein thrombosis (DVT), where blood clots form in deep veins, typically in the legs, and pulmonary embolism (PE), where a clot travels to the lungs, which can be life-threatening.

Crucially, for individuals identified as being at elevated risk, CardiOmicScore was capable of flagging this heightened cardiovascular risk up to 15 years before the emergence of any clinical symptoms. This extended predictive window is unprecedented and opens up a vast new landscape for preventive healthcare.

A Paradigm Shift Towards Proactive Healthcare

The development of CardiOmicScore exemplifies a broader, transformative shift occurring within the field of precision medicine. Historically, medical approaches have often been reactive, focusing on treating diseases once they have taken hold. Traditional genetic risk assessments, while valuable for understanding inherited predispositions, provide a relatively static view of an individual’s health trajectory. In contrast, multiomics-based tools like CardiOmicScore offer a far more dynamic and nuanced assessment by continuously monitoring biological signals that evolve over time.

The implications of this advancement are profound. In the not-too-distant future, a simple blood draw could potentially yield a comprehensive risk profile encompassing multiple cardiovascular diseases simultaneously. This detailed information would empower both patients and healthcare providers with an extended timeframe to implement proactive strategies. These strategies could include targeted lifestyle modifications, such as dietary adjustments or increased physical activity; more frequent and personalized health monitoring; or the early initiation of preventive therapies, tailored to an individual’s specific risk factors.

Professor Zhang emphasized the ultimate goal of this research: "We aim to leverage technology to identify and prevent diseases before they develop," he stated. "By shifting health management from reactive treatment to proactive prediction and intervention, we aim to create a lasting impact for both public health and individual patient care." This vision represents a fundamental reorientation of healthcare, moving away from managing illness towards actively promoting wellness and preventing disease at its earliest stages.

Broader Impact and Future Directions

The success of CardiOmicScore is not merely a triumph for HKUMed but a beacon of hope for global cardiovascular health. The increasing prevalence of CVDs, fueled by factors such as aging populations, lifestyle changes, and environmental influences, necessitates innovative solutions. This AI-driven multiomics approach offers a scalable and potentially cost-effective method for widespread risk stratification, particularly in resource-constrained settings where traditional advanced diagnostics may be less accessible.

Further research will likely focus on validating CardiOmicScore in diverse populations and exploring its integration into routine clinical workflows. Efforts may also be directed towards refining the AI model to predict a wider range of diseases or to identify specific subtypes of cardiovascular conditions with even greater precision. The potential for continuous monitoring, where individuals undergo periodic blood tests to track changes in their CardiOmicScore, could provide real-time insights into their evolving health status, allowing for dynamic adjustments in preventive strategies.

The development team, led by Professor Zhang Qingpeng from the Department of Pharmacology and Pharmacy at HKUMed and the HKU Musketeers Foundation Institute of Data Science (IDS), with Luo Yan as the first author from the HKU IDS, has laid a robust foundation for a new era of predictive and preventive medicine. As the understanding of complex biological systems deepens and AI capabilities continue to advance, tools like CardiOmicScore are poised to redefine how we approach cardiovascular health, ultimately aiming to reduce the burden of disease and enhance the quality of life for millions worldwide. The transition from treating the symptoms of heart disease to preventing its very onset marks a monumental stride in medical science, driven by the power of artificial intelligence and a profound understanding of human biology.