Researchers at the LKS Faculty of Medicine at the University of Hong Kong (HKUMed) have developed an artificial intelligence-based tool that could help predict serious cardiovascular diseases many years before symptoms appear. The system, called “CardiOmicScore,” uses information from a single blood test to estimate a person’s future risk for six major cardiovascular diseases (CVDs): coronary heart disease, stroke, heart failure, atrial fibrillation, peripheral arterial disease, and venous thromboembolism. In individuals at increased risk, the model was able to detect warning signs up to 15 years before clinical onset. The findings were published in *Nature Communications *.
A Blood Test that Assesses Current Health Status
Cardiovascular diseases remain the leading cause of death worldwide and were responsible for approximately 19.8 million deaths in 2022 alone. This means that about one in three deaths worldwide is attributable to this group of diseases. Experts believe that a significant proportion of cardiovascular diseases could be prevented or at least delayed through preventive measures. These include a balanced diet, regular physical activity, not smoking, monitoring blood pressure, blood sugar, and cholesterol, as well as the early identification of individual risk profiles. Against this backdrop, new diagnostic methods capable of detecting biological changes long before the first symptoms appear are becoming increasingly important.
Doctors typically assess cardiovascular risk based on factors such as age, blood pressure, smoking status, and other standard clinical measurements. While these indicators are useful, they may not be able to detect the earliest biological changes occurring in the body before a disease becomes apparent. As a result, some people may not be identified as high-risk patients until the best opportunities for prevention have already begun to narrow. Genetic risk tests offer another way to estimate the likelihood that a person will develop a disease. Polygenic risk scores, for example, combine the effects of many genetic variants into a single measure of inherited risk. However, a person’s genetic makeup is largely determined at birth. This means that genetic scores cannot fully reflect more immediate changes caused by diet, exercise, aging, disease, environmental influences, or other health-related factors. CardiOmicScore was developed to provide a more up-to-date picture of what is happening in the body.
AI Combines Thousands of Biological Signals
To develop the tool, the HKUMed team used deep learning to combine multiple levels of biological information. This approach is known as multiomics, as it brings together data from various fields of biology, including genomics, metabolomics, and proteomics. Genomics examines genetic information. Proteomics focuses on proteins, which perform many essential functions in the body. Metabolomics examines small molecules, known as metabolites, that are produced when the body processes food, generates energy, and responds to disease. The researchers analyzed large-scale population data from the UK Biobank. Their model examined 2,920 circulating proteins and 168 metabolites measured in blood samples.
Together, these molecules can provide a detailed snapshot of a person’s current biological state. They can reflect subtle changes in immune activity, metabolism, and vascular health even before noticeable symptoms appear. Professor Zhang Qingpeng, associate professor at the Department of Pharmacology and Pharmacy at HKUMed, explained: “Genes determine where we start—they define our baseline health risk. Proteins and metabolites, however, reflect our current physical health. Our AI tool was developed to decipher these complex molecular signals so that doctors and patients can identify risks much earlier, which may alter the course of the disease through timely lifestyle changes and early prevention.”
Prediction of Six Cardiovascular Diseases
The study’s results show that the CardiOmicScore can convert complex molecular information from a single blood sample into an individualized assessment of long-term cardiovascular risk. In doing so, the AI model proved to be significantly more powerful than conventional polygenic risk scores, which are based exclusively on genetic information. While genetic tests merely reflect inherited susceptibility to certain diseases, the CardiOmicScore also takes into account the body’s current biological state. This makes it possible to detect disease-relevant changes influenced by aging processes, diet, physical activity, inflammation, or other environmental and lifestyle factors.
The model’s predictive accuracy was further improved when classic clinical information—such as age, gender, and other basic medical data—was incorporated into the calculation. This demonstrates that the combination of molecular biomarkers and established risk factors enables a particularly precise assessment of an individual’s disease risk. The system was specifically developed to predict the risk of six of the world’s most significant cardiovascular diseases:
- Coronary heart disease (CHD): Deposits in the coronary arteries restrict the blood supply to the heart muscle. This disease is the most common cause of heart attacks and ranks among the leading causes of death worldwide.
- Stroke: A stroke occurs either due to the blockage of a blood vessel supplying the brain or, less commonly, due to a brain hemorrhage. It can lead to permanent neurological damage within minutes and is one of the most common causes of disability in adulthood.
- Heart Failure: In heart failure, the heart is no longer able to pump enough blood throughout the body. Those affected often suffer from shortness of breath, rapid fatigue, reduced physical performance, and fluid retention.
- Atrial fibrillation: This most common type of persistent cardiac arrhythmia results in uncoordinated electrical activity in the atria. This can lead to the formation of blood clots, which significantly increase the risk of stroke. In addition, the risk of heart failure and other cardiovascular complications rises.
- Peripheral arterial disease (PAD): In PAD, arteries—usually in the legs—narrow or become blocked due to atherosclerosis. Typical symptoms include exertion-induced pain in the calves (“intermittent claudication”); in advanced stages, slow-healing wounds or even tissue loss may occur.
- Venous thromboembolism (VTE): This includes, in particular, deep vein thrombosis and pulmonary embolism. Blood clots usually form in the deep veins of the legs and can break loose, travel through the bloodstream to the lungs, and cause life-threatening circulatory disorders there.
It is particularly noteworthy that the CardiOmicScore was able to detect corresponding warning signs in individuals with an elevated risk of disease up to 15 years before the onset of clinical symptoms. This opens up the possibility of identifying at-risk individuals much earlier than with previous diagnostic methods. Such early risk assessment could give doctors more time to initiate preventive measures such as lifestyle changes, frequent follow-up examinations, or, if necessary, medication. The goal is to prevent the onset of serious cardiovascular diseases as much as possible—or at least to significantly delay it.
From Treatment to Early Prevention
The development of the CardiOmicScore reflects the shift toward precision medicine in healthcare. While genetic risk scores primarily reflect inherited susceptibility to diseases, multi-omic analyses provide an up-to-date picture of a person’s biological health status. Since proteins and metabolites change with age, lifestyle, or disease, they can provide early indications of health risks that genetic tests alone cannot detect.
In the future, a single blood sample could be sufficient to create a comprehensive risk profile for multiple cardiovascular diseases simultaneously. This would give doctors more time to initiate targeted preventive measures—such as lifestyle changes, close monitoring, or early medication—before the first symptoms appear. However, further studies are needed before this approach can be widely adopted in clinical practice. The method must be validated across different population groups, and its actual benefit for the prevention of severe cardiovascular diseases must be demonstrated.
Professor Zhang explained: “Our goal is to use technology to detect and prevent diseases before they develop. By shifting health management from reactive treatment to proactive prediction and intervention, we aim to achieve a sustainable impact on both public health and individual patient care.”