My research focuses on the development and evaluation of machine learning and statistical methods for healthcare. My main interests include multimodal machine learning, clinical risk prediction, causal machine learning, and physiological signal and medical image processing. I have also been the lead biostatistician on a variety of medical research projects in fields ranging from diabetes pharmacoepidemiology to orthopaedic surgery.
My MSc research investigated methods for detecting and quantifying nociception during general anaesthesia using single-channel electroencephalography. The project included EEG signal processing, unsupervised machine learning, supervised machine learning, and the development of a novel nonlinear mixed-effects model combining machine learning with pharmacodynamic modelling techniques. The thesis is available on the University of Waikato research commons.
This dissertation applied conditional average treatment-effect estimators to post-anaesthetic recovery data. I used explainability methods to characterise treatment-effect heterogeneity and investigate how these models could inform patient selection for Advanced Recovery Room Care. Findings from the wider project were presented at the 2025 ANZCA Annual Scientific Meeting.
My medical elective project developed a multimodal machine-learning model for predicting the risk of primary open-angle glaucoma. This work was completed as an honours-equivalent dissertation at the University of Auckland and presented at the 2022 Glaucoma New Zealand Professional Symposium.
I have contributed statistical analyses to research in diabetes, orthopaedic surgery, medication use, and health-service delivery. This work has included data cleaning and linkage, regression and mixed-effects modelling, algorithm evaluation, longitudinal analysis, and the preparation of journal manuscripts, reports, conference presentations, and posters.
At the 2023 Hack Aotearoa healthcare datathon, I contributed to a project developing models to predict multidrug-resistant organisms from clinical data. The project received the awards for Best Overall and Most Reproducible.
This project, which I undertook for my medical selective, involved processing and analysing EEG and physiological data, visualising multidimensional signals, and developing a mixed-effects explanatory model to investigate responses during surgery.
I contributed to establishing the first New Zealand adult site for the Australian Diabetes Data Network. The project included clinical-data cleaning and analysis, as well as evaluation of algorithms for differentiating type 1 and type 2 diabetes using routinely collected data.
Further details and research outputs are available through my Google Scholar profile.