Cardiovascular disease remains the leading cause of death worldwide, responsible for approximately 19.8 million deaths in 2022 alone. Despite decades of research and improved treatments, one of the biggest challenges in cardiology hasn’t changed: by the time symptoms appear, the disease has often already been developing silently for years.
Researchers at the LKS Faculty of Medicine of the University of Hong Kong (HKUMed) just unveiled an AI-powered tool that could dramatically change that timeline. Published in Nature Communications, the new system — called CardiOmicScore — can estimate a person’s risk of six major cardiovascular diseases using just a single blood test, flagging warning signs up to 15 years before clinical symptoms emerge.
Why Current Risk Assessment Falls Short
Doctors currently assess cardiovascular risk primarily through standard clinical measurements — age, blood pressure, cholesterol levels, smoking history, and similar factors. These indicators are genuinely useful, but they have an important limitation: they often don’t capture the earliest biological changes happening inside the body long before a disease becomes clinically apparent.
Nanoplastics In Drinking Water May Be Making Dangerous Bacteria Stronger
As a result, some people aren’t flagged as high-risk until much of the optimal window for prevention has already narrowed.
Genetic risk tests offer an alternative approach. Polygenic risk scores combine the effects of many genetic variants into a single measure of inherited risk. The problem is that a person’s genetic makeup is largely fixed at birth — meaning these scores cannot reflect the more immediate changes caused by diet, exercise, aging, illness, environmental exposures, or evolving lifestyle factors that meaningfully influence real-world health over time.
CardiOmicScore was specifically designed to close this gap — providing a picture of what’s happening in the body right now, rather than relying solely on inherited risk that never changes.
How CardiOmicScore Actually Works
To build the tool, the HKUMed research team used deep learning to combine multiple layers of biological data — an approach known as multiomics, which integrates information from several distinct areas of biology simultaneously:
- Genomics — genetic information
- Proteomics — the study of proteins, which carry out most of the body’s essential biological functions
- Metabolomics — the study of small molecules called metabolites, produced as the body processes food, generates energy, and responds to disease and stress
Using large-scale population data from the UK Biobank, the researchers analyzed 2,920 circulating proteins and 168 metabolites measured directly from blood samples.
Together, these thousands of biological signals create a remarkably detailed snapshot of a person’s current physiological state — potentially reflecting subtle shifts in immune activity, metabolism, and vascular health long before any symptoms become noticeable.
“Genes determine where we start — they define our baseline health risk,” explained Professor Zhang Qingpeng, Associate Professor in the Department of Pharmacology and Pharmacy at HKUMed and the study’s lead researcher. “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.”
Predicting Six Major Cardiovascular Diseases At Once
One of the most significant features of CardiOmicScore is its scope. Rather than assessing risk for a single condition, the tool was designed to evaluate risk across six major cardiovascular diseases simultaneously:
- Coronary artery disease — narrowing of the arteries supplying the heart
- Stroke — interrupted blood flow to the brain
- Heart failure — the heart’s reduced ability to pump blood effectively
- Atrial fibrillation — an irregular heartbeat that increases stroke risk and other complications
- Peripheral artery disease — narrowed blood vessels that reduce circulation to the limbs
- Venous thromboembolism — dangerous blood clots forming in a vein that may travel to the lungs
The results demonstrated that CardiOmicScore performed substantially better than conventional polygenic (genetic) risk scores across these conditions. Its predictive accuracy improved even further when researchers incorporated basic clinical information such as age and gender alongside the molecular data.
Among individuals identified as high-risk, the system successfully flagged elevated cardiovascular risk up to 15 years before clinical symptoms actually emerged — a remarkably early warning window that could fundamentally change how preventive cardiology is practiced.
From Reactive Treatment To Proactive Prevention
This research reflects a broader and increasingly important shift happening across precision medicine more generally: moving away from reacting to disease after it appears, and toward genuinely predicting and preventing it well in advance.
Traditional genetic approaches provide a relatively fixed estimate of lifelong inherited risk. Multiomics tools like CardiOmicScore, by contrast, offer a dynamic assessment — one that can track biological signals that change meaningfully over the course of a person’s life, in response to lifestyle, environment, and aging.
Looking ahead, the research team envisions a future where a single small blood sample could generate a detailed, comprehensive risk profile covering multiple cardiovascular diseases at once — giving both patients and doctors significantly more lead time to respond through targeted lifestyle changes, closer clinical monitoring, or other preventive interventions, well before disease onset.
“We aim to leverage technology to identify and prevent diseases before they develop,” Professor Zhang said. “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.”
What This Could Mean For Patients And Doctors
If validated further and eventually integrated into routine clinical practice, tools like CardiOmicScore could reshape preventive cardiology in several meaningful ways:
- Earlier identification of high-risk individuals who would otherwise appear low-risk under conventional assessment methods
- More time for lifestyle intervention — diet, exercise, smoking cessation, and other preventive strategies are generally more effective the earlier they begin
- More efficient use of clinical monitoring resources, directing closer follow-up toward those who genuinely need it most, based on current biological signals rather than static inherited risk alone
- A more comprehensive risk picture, since assessing six conditions simultaneously from one test is significantly more efficient than separate assessments for each disease
Important Context And Next Steps
This research represents a significant scientific advance, but it’s still important to understand its current stage. The study was conducted using large-scale population data from the UK Biobank, and while the results are compelling, broader validation across more diverse populations, further clinical testing, and regulatory approval processes would typically be needed before a tool like this becomes a standard, widely available clinical test.
The research team, led by Professor Zhang Qingpeng alongside first author Luo Yan from the HKU Musketeers Foundation Institute of Data Science, continues to refine and validate the CardiOmicScore system as part of the broader push toward proactive, prediction-based precision medicine.
Key Takeaways
- Researchers at the University of Hong Kong developed CardiOmicScore, an AI tool that predicts six major cardiovascular diseases from a single blood test
- The tool analyzes 2,920 proteins and 168 metabolites, capturing current biological health rather than fixed genetic risk
- CardiOmicScore flagged elevated cardiovascular risk up to 15 years before clinical symptoms emerged in high-risk individuals
- The system significantly outperformed traditional genetic (polygenic) risk scores
- This research reflects a broader shift in medicine toward proactive prediction and early prevention, rather than reactive treatment after disease onset
Source: The University of Hong Kong — July 19, 2026
Journal Reference: Yan Luo, Nan Zhang, Jiannan Yang, Mengyao Cui, Kelvin K. F. Tsoi, Gregory Y. H. Lip, Tong Liu, Qingpeng Zhang. AI-based multiomics profiling reveals complementary omics contributions to personalized prediction of cardiovascular disease. Nature Communications, 2026; 17 (1).
DOI: 10.1038/s41467-026-68956-6
