Home/Research and AI

Eighteen years of recording what happened. Now researching what may happen next.

After eighteen years of recording what happened, a national cardiac device registry starts to show what may happen next. HRI's research draws on longitudinal cardiac device records captured through routine care across multiple manufacturers, hospitals and years within Ireland's National Cardiac Device Registry.

Horizon Europe

CARAMEL

HRI is a partner in CARAMEL, a Horizon Europe project to improve cardiovascular health for women aged 40 to 60, particularly during perimenopause and menopause, with partners including Tampere University. HRI provides research outputs derived from Ireland's National Cardiac Device Registry. Detailed results belong to the consortium and are published through its deliverables.

caramel-project.eu

Basel, 12 to 13 November 2026

ESC Digital & AI Summit

An abstract on HRI's late-fusion AI framework has been accepted for presentation at the European Society of Cardiology's Digital & AI Summit 2026. Results will be published here once they have been presented.

Belfast, cross-border

PEACEPLUS Health Frontiers

HRI is an industry partner in the Health Frontiers Technology Innovation Centre, led by Ulster University with €9.1 million in PEACEPLUS funding, within its Connected Health and Wellbeing Cluster.

Read about HRI’s involvement

This is a research programme. The models described are not a feature of HRI Pro, are not used in clinical decision-making and are not medical devices.

How the research is done

The modelling comes to the data. The data does not leave.

Inside HRI's own environment

Within CARAMEL, AI modelling is applied to longitudinal device data inside HRI's own environment, under the project's data governance and ethical framework. What is shared with the consortium is research output, not patient records.

Governance enforced by architecture

A permit that says data may not be extracted is a contract. A system where extraction is technically blocked is an architecture. HRI argues for the second: computation travels to the data, and results leave through output checking.

Structured, governed, longitudinal

AI-ready health data is structured at the point of capture, governed lawfully and transparently, and follows the patient across years, institutions and device changes. It is extremely difficult to recreate retrospectively from fragmented PDFs, free text and disconnected systems.

Longitudinal. National-scale. Built over 18 years.

Manufacturer specifications say what a device is designed to do. Trial data says how it performed in a selected population for a defined period. A manufacturer's post-market data covers only its own devices. A registry built through routine care sees devices from multiple manufacturers side by side, in the same hospitals, followed over the same years.

Battery longevity in practice

Degradation curves that vary by device model, programming, pacing burden and patient physiology, measured across recorded follow-ups rather than estimated at design.

Lead performance over a decade and more

Lead survival at five, ten and fifteen years, and trends that may signal a problem long before it is clinically apparent.

Patterns associated with earlier intervention

Research into whether longitudinal device and follow-up patterns can identify patient groups associated with earlier clinical intervention.

Evidence that starts on day one

When Ireland's first dual-chamber leadless pacing system was completed at the Mater in April 2026, it was captured on the registry during the procedure. A new device class needs longitudinal evidence from the moment it enters practice, not a retrospective audit years later.

Read how it was recorded

How we approach it

Research sits beside the clinical platform, not inside it. Anything that might one day inform care would first have to go through validation, regulatory assessment and the governance of the hospitals whose patients the data describes. The EU Medical Device Regulation and the European Health Data Space both point towards routine-care data being usable for research under proper governance. National device registries sit where the two meet.

The data starts with the clinical platform.

See how the record is built, stage by stage, on demonstration data.