Research area

AI and Physiological Data

We develop interpretable models for physiological signals and longitudinal clinical data, with emphasis on reliability, clinical meaning and reproducibility.

01 — Clinical care

Ventilator waveform intelligence

Real-time acquisition, breath segmentation and identification of trigger delay, ineffective effort, double triggering and flow limitation.

Our lab is working on developing noninvasive devices to extract respiratory parameters to determine lung health. These devices will not require additional effort from patients, making them ideal for children, the elderly and individuals with physical limitations.

02 — Longitudinal health

Diabetes progression forecasting

Prediction of later diabetes or metabolic progression using repeated cohort measurements, risk factors and longitudinal state information.

High-risk factors for diabetes can be identified through advanced machine learning techniques, particularly clustering algorithms applied to longitudinal cohort data. This method groups individuals based on similar health characteristics, facilitating the identification of phenotypes associated with increased diabetes risk.

03 — Cardiac signals

Interpretable ECG morphology

This research addresses the challenge of ECG interpretation variability by developing a mathematical model of a standardised ECG signal.

The model, validated against ECGs from various arrhythmias, allows for parameter estimation to quantify deviations from normal. These parameters can then be used in machine learning models to predict arrhythmias, potentially enabling earlier detection and improved patient outcomes.

How we work

Methodological priorities

Four commitments that shape how every model in the lab is built, tested and reported.

Data quality

Acquisition, pre-processing, artefact detection and quality control are treated as part of the scientific method.

Interpretable outputs

Models are linked to physiological features and clinically meaningful events wherever possible.

External validation

Performance is evaluated across participants, devices, centres or sites when the data permit.

Reproducible analysis

Open, specific code, traceable inputs and appropriate acceptance criteria are integral to each study.

Explore Other Research Areas

Join the lab

Explore a project with us

We welcome students and collaborators working in clinical AI, signal processing and biomedical data science.

View open positions