A University of Hong Kong research team develops AI tool that can turn a single blood test into an early warning system for heart diseases up to 15 years before symptoms appear.
A research team at the University of Hong Kong has developed an artificial intelligence tool that can turn a single blood test into an early warning system for six major cardiovascular diseases, potentially up to 15 years before symptoms appear.
The system, called CardiOmicScore, uses deep learning to analyze thousands of molecules in the blood and translate them into a personalized risk score. The study, published in the journal Nature Communications, is led by Zhang Qingpeng, an associate professor in the Department of Pharmacology and Pharmacy at the LKS Faculty of Medicine of the University of Hong Kong (HKUMed) and a member of the HKU Musketeers Foundation Institute of Data Science.
Cardiovascular diseases remain the world’s leading killer, responsible for an estimated 19.8 million deaths in 2022 alone. Doctors typically assess a person’s heart risk using factors such as age, blood pressure and smoking status. Those measures are helpful, but they often miss the earliest biological changes that signal trouble years before a heart attack, stroke or other event.
In recent years, polygenic risk scores — which estimate disease risk based on inherited genetic variants — have gained attention. But genes are fixed at birth and do not change with lifestyle or environment. That means genetic scores cannot fully capture how current habits, diet, stress or pollution are affecting someone’s heart and blood vessels right now.
The HKUMed team set out to fill that gap by building a tool that reflects the body’s real-time health status.
To do this, they turned to “multiomics,” an approach that combines different layers of biological data. Using large-scale population data from the UK Biobank, the researchers applied deep learning techniques to integrate genomics (DNA information), proteomics (proteins circulating in the blood) and metabolomics (small molecules involved in metabolism).
The study analyzed 2,920 circulating proteins and 168 metabolites measured from blood samples. These molecules act like real-time recorders of the body, capturing subtle shifts in the immune system, metabolism and vascular health that may precede disease by many years.
CardiOmicScore takes these complex molecular patterns and converts them into individualized risk scores for six common cardiovascular conditions: coronary artery disease, stroke, heart failure, atrial fibrillation, peripheral artery disease and venous thromboembolism.
According to the researchers, the tool substantially outperformed conventional polygenic risk scores in predicting who would go on to develop these conditions. When the AI-generated scores were combined with basic clinical information such as age and gender, prediction accuracy improved even further.
Crucially, the system was able to flag elevated risk as far as 15 years before clinical onset in high-risk groups. That kind of lead time could give patients and doctors a much larger window to act — through lifestyle changes, closer monitoring or preventive treatments — before irreversible damage occurs.
The work signals a broader shift in precision medicine. For years, much of the focus has been on genetics: reading a person’s DNA to estimate their lifetime risk of disease. The HKUMed study points toward a more dynamic model that layers genetic predisposition with real-time molecular snapshots of how the body is functioning today.
“Genes determine where we start—they define our baseline health risk. 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,” Zhang said in a news release.
In the future, the researchers envision that a small-volume blood sample could be enough to generate a comprehensive cardiovascular risk profile across multiple diseases in a single test. That could change how routine checkups are done, especially for people in midlife or with known risk factors.
Instead of waiting for symptoms like chest pain or shortness of breath, clinicians could use tools like CardiOmicScore to identify silent but rising risk and intervene earlier. For public health systems, such an approach could help target prevention efforts more efficiently and potentially reduce the burden of heart attacks, strokes and related conditions.
“We aim to leverage technology to identify and prevent diseases before they develop. 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.” Zhang added.
While the findings are promising, further work will be needed before CardiOmicScore can be widely used in clinics. That includes validating the tool in more diverse populations, integrating it into health care workflows and assessing how acting on its predictions affects long-term outcomes.
Still, the research offers a glimpse of a future in which a routine blood draw could do far more than check cholesterol or blood sugar. By decoding thousands of molecular signals at once, AI-driven tools may help move heart care from reacting to crises to preventing them years in advance.
Source: The University of Hong Kong