Title : Behavioural signals before clinical thresholds: How AI Is rewriting early cardiovascular risk

Cardiovascular disease remains the leading cause of death worldwide, yet most patients are still identified only after they have crossed a clinical threshold: a hypertensive reading, an abnormal lipid panel, a stress test that finally turns positive. By the time these markers cross diagnostic cutoffs, the underlying biology has been developing for years, and so has the behaviour driving it. The most consequential signals tend to appear long before the numbers do.

This talk examines what becomes possible when we stop treating behaviour as a soft variable and begin reading it as the earliest available indicator of cardiovascular risk. Sleep fragmentation, declining step counts, shifts in resting heart rate variability, missed prescription refills, changes in daily routines, and patterns of social disengagement all carry predictive weight. Machine learning models trained on these passive and semi-passive signals are beginning to flag at-risk patients months, sometimes years, ahead of standard screening, with growing evidence from peer-reviewed trials and real-world primary care deployments.

The argument is not that AI replaces clinical judgment. It is that combining behavioural data, contextual intelligence, and predictive modelling allows us to see risk earlier, intervene with smaller and more humane changes, and reach the people most often missed by population screening, including women, younger adults, and patients in lower-resource settings. Behavioural economics adds the layer that pure prediction cannot supply: an understanding of why people defer care, why adherence collapses, and how environments shape the choices a patient is actually able to make.

The session brings together findings from current behavioural economics research, recent work on digital biomarkers in cardiology, and lessons from early deployments of risk models in clinical practice. It closes with an honest look at the limits of this approach, including data quality, equity, consent, model drift, and the risk of medicalising ordinary life. The goal is a clearer picture of what it means to act before the threshold is crossed, and what cardiology stands to gain from listening earlier.

Keywords: Behavioural economics; cardiovascular risk prediction; digital biomarkers; machine learning in cardiology; preventive cardiology; behaviour change; early detection; passive sensing; health equity; clinical decision support; primary prevention; AI in healthcare.

Dr Iris-Panagiota Efthymiou is a Senior Lecturer at the Institute for Study Abroad, in University of Roehampton (Futurelearn) and RCL. She has also taught at the University of Greenwich, the University of East London, and other European Universities. With more than 20 years of experience in entrepreneurship, international affairs, and executive leadership, she combines academic expertise with global policy engagement. She has delivered over 400 keynote speeches across 30 countries on four continents, addressing international institutions, governments, and academic forums. Her policy influence includes work with UK parliamentarians through the All-Party Parliamentary Group on AI, as well as collaborations with diplomats, business leaders, and international organisations. She has served as a college director, founded a public affairs consultancy, and spoken at the United Nations Headquarters in Geneva. She is a Board Member of HAPSc and a Member of the Academic Board of LabHEM at the University of Piraeus, Greece.. Dr Efthymiou has authored 20 books, more than 60 peer-reviewed articles and book chapters, and serves as Chief Editor of the Journal of Politics and Ethics in New Technologies and AI. Her research and public work focus on behavioral economics, AI ethics, healthcare, and public policy. She holds a PhD in Behavioral Economics, a Master’s in Health Economics and Management, and a Bachelor’s in Economics. Her work consistently advocates for transparency, inclusivity, and fairness in shaping technology, healthcare, and the future of work.