Researchers have developed a new artificial intelligence (AI) model that could significantly improve how heart conditions are detected, monitored and managed — whether in intensive care units or in people’s own homes.
Heart disease remains one of the leading causes of death worldwide. Clinicians rely on heart signals such as electrocardiograms (ECGs), which record electrical activity, and photoplethysmograms (PPGs), which track blood flow using light sensors found in many wearable devices.
However, these signals are collected in very different ways, depending on the setting, device and level of care.
Most existing AI tools are designed for a single type of heart signal or a specific task. This limits their usefulness in real‑world healthcare, where data are often incomplete, inconsistent or gathered using different technologies.
To address this problem, the researchers developed a ‘foundation model’ for cardiac monitoring, known as the cardiac sensing foundation model, or CSFM. Instead of being built for one purpose, CSFM is trained on large and diverse datasets so it can adapt to many tasks and data types.
The model was trained using heart signals and clinical reports from around 1.7 million people, combining ECGs, PPGs and associated medical text.
Dr Xiao Gu, Senior Research Associate in Medical AI at the University’s Institute of Biomedical Engineering (IBME) and lead author, said: “By learning common patterns across these varied data, CSFM can analyse heart signals from different devices and settings without needing to be redesigned each time.
“In our testing, CSFM consistently outperformed traditional models across a wide range of applications. These included diagnosing heart disease, estimating age and body mass index, measuring blood pressure or predicting clinical outcomes such as mortality risk. Significantly, the model performed well with both hospital‑grade 12‑lead ECGs and simpler single‑lead or wearable devices, and was equally adept at handling ECG data alone, PPG data alone or a combination of both.
“This flexibility makes it particularly promising for use in community care and in regions with limited access to advanced medical equipment.”
The research, which was supported by the NIHR Oxford Biomedical Research Centre, was published in Nature Machine Intelligence, with the paper selected as the cover article for the journal’s February issue
Co-author Professor David Clifton, Royal Academy of Engineering Chair of Clinical Machine Learning in the Department of Engineering Science and NIHR Research Professor, added: “Heart disease is the leading cause of death, yet many patients with the condition lack access to consistent monitoring. Xiao has led the development of a new ‘foundation model’ for heart health – imagine a large language model like ChatGPT, but where the ‘language’ is healthcare sensor data. This model was built using data from 1.7 million patients, one of the world’s largest datasets of its kind. It then learns patterns between sensors, types of patients and healthcare setting, such as hospital, home etc.
“The result is extremely promising for helping doctors detect heart problems earlier – and more fairly, such as in our longstanding collaborations in low-income countries, including the Oxford University Clinical Research Units in Vietnam and Nepal. This breakthrough could bring reliable heart monitoring to more people, including those in remote or under-served places. I think it very well-deserved that this fantastic work led by Xiao has been given the accolade of the cover feature for the primary Nature journal in AI.”
As well as the Oxford researchers, the research team included experts from Brazil, Hong Kong and Sweden, as well as Imperial College London and the University of Nottingham in the UK.