Davood Roshan is a Lecturer in Statistical Science in the School of Mathematical and Statistical Sciences at the University of Galway, and a Principal Investigator at CÚRAM, the Research Ireland Centre for Medical Devices. He completed his BSc and MSc in Statistics at the University of Tehran, Iran, in 2009 and 2013, followed by a PhD in Biostatistics at the University of Galway in 2020. Before his permanent appointment as Lecturer in 2022, he held several fixed-term academic and research roles, including lecturer and postdoctoral positions.

Davood’s research is in Statistics and Data Science, with a focus on developing and applying statistical methods for complex biomedical, clinical, longitudinal and sensor-based data. His work is concerned with turning complex data into interpretable evidence for monitoring, prediction and decision-making, particularly in health, sport science and other applied settings. A central part of his research focuses on identifying meaningful change over time, developing personalised monitoring approaches, and translating statistical model outputs into tools that can be used by clinicians, researchers and other non-statistical users.

Alongside his methodological research, Davood is actively involved in collaborative and enabling statistical research across oncology, cardiovascular health, diabetes, sport science, concussion monitoring, wound care, dermatology, digital health, environmental monitoring and biomedical imaging. He has secured and contributed to successful research funding of more than €1.24 million, and has supervised PhD students and postdoctoral researchers across several areas of applied and translational statistics.

Davood’s primary research interests are in applied and translational statistics, with a focus on developing statistical methods that turn complex data into interpretable evidence for decision-making. One major area of his work is anomaly detection in longitudinal monitoring. This research focuses on identifying meaningful changes in individual trajectories when data are noisy, repeated over time, and influenced by personal characteristics, context and measurement uncertainty.

A second area of Davood’s research is translational statistics. This work focuses on making statistical models more understandable, accessible and useful for clinicians, applied researchers and other non-statistical users. It includes the development of interactive tools, visualisation methods and open-source R packages that help communicate model outputs and support evidence-based decision-making.

A third area of his research is predictive modelling for personalised medicine. This work focuses on integrating high-dimensional and multimodal biomedical data, including ECG signals, clinical and lifestyle factors, genomic data, spatial transcriptomics and whole slide histopathology images. The aim is to improve prediction, risk stratification and clinical interpretation of patient outcomes.

Davood also contributes to collaborative statistical research across health, sport science, biomedical science, environmental monitoring and data science. In these projects, his role is to provide statistical expertise in study design, data analysis, interpretation and the generation of reliable evidence from complex real-world data.

Statistical Modelling

Statistical Computing

Predictive Modelling

Data Science

Probability

Multivariate Adaptive Reference Regions for Longitudinal Data Monitoring
PhD project supervised by Davood Roshan; PhD student: Forough Pazhuheian (Funded by College of Science and Engineering Scholarship)

This PhD project develops multivariate adaptive reference regions for personalised monitoring of longitudinal biomarker data. The work extends adaptive reference range methods from single biomarkers to panels of related biomarkers, allowing joint monitoring of several measurements while controlling false alarms. The project uses multivariate mixed-effects models to account for biomarker dependence, individual variability and covariate effects, with the aim of developing efficient methods that update as new patient data become available.

Predicting Peripheral Artery Disease Outcomes Using ECG, Lifestyle and Clinical Factors
PhD project supervised by Davood Roshan; PhD student: George Aryee (Funded by Hardiman Scholarship)

This PhD project focuses on improving prediction of long-term outcomes in peripheral artery disease (PAD), including hospitalisation and mortality. PAD occurs when blood vessels in the legs and feet become narrowed or blocked, reducing blood flow and increasing the risk of poor wound healing, amputation, heart attack, stroke and death. Current prediction approaches rely mainly on lifestyle and clinical factors such as smoking, alcohol intake, diet, cholesterol, hypertension and kidney function. This project investigates whether adding ECG-derived features can improve risk prediction and provide a more complete profile of patient risk. By combining ECG signals with lifestyle and clinical information from a large cohort of PAD patients, the work aims to identify factors associated with poor outcomes and support earlier identification of patients who may benefit from timely intervention. The project has potential to inform personalised prevention strategies, improve clinical decision-making and reduce the burden of advanced PAD.

AI-Based Prediction of Gene Expression from Histopathology and Whole Slide Images
Research project supervised by Davood Roshan; Postdoctoral researcher: Asfand Yaar (funded by MedTrain+ Marie Skłodowska-Curie COFUND)

This postdoctoral project develops AI-based methods to predict gene expression from routine H&E-stained histopathology images, with a focus on ovarian cancer. Existing models linking whole slide images to molecular data often rely on bulk sequencing, which can miss important spatial variation within tissue. This project addresses this limitation by developing deep learning approaches that estimate localised gene expression profiles directly from whole slide images, helping to capture spatial heterogeneity in tumour biology. The work aims to bridge histopathology and molecular profiling, providing a faster, less invasive and more cost-effective route to advanced cancer diagnostics. By combining model development with assessment of clinical utility, scalability and integration into diagnostic workflows, the project supports more accessible precision oncology and personalised treatment decision-making.

Concussion Monitoring and Normative Ranges in Irish Horseracing
Research project supervised by Siobhan O’Connor (DCU) and Davood Roshan; Postdoctoral Researcher: Maqsood Hossein Shah (funded by Health Research Board Secondary Data Analysis Project)

This postdoctoral project examines concussion risk, presentation and clinical outcomes in Irish horseracing using routinely collected demographic, performance and clinical assessment data from 2009–2023. The project addresses an important evidence gap, as there is limited horseracing-specific concussion research internationally despite the high-risk nature of the sport and the occupational demands placed on jockeys. The work has four linked aims: to estimate concussion incidence and outcomes in Irish horseracing; to develop jockey-specific normative data for baseline concussion assessments; to create personalised reference ranges from longitudinal clinical assessments as an early warning system; and to identify risk factors for concussion and persistent post-concussion symptoms. By working with clinical and regulatory partners, including the Irish Horseracing Regulatory Board and the Sports Surgery Clinic, the project aims to generate evidence that can improve concussion prevention, assessment and management in horseracing nationally and internationally.

Data-Driven Modelling of Jockey Falls, Injuries and Horse Fatalities in Irish Horseracing
PhD project supervised by SarahJane Cullen (DCU), Siobhán O’Connor (DCU) and Davood Roshan; PhD student: Hamza Qadeer (Co-funded by Dublin City University and Jockey Accident Funds)

This PhD project uses anonymised secondary data from the Irish Horseracing Regulatory Board, Horse Racing Ireland and Race IQ to model risk factors for jockey falls, injuries, fractures and horse fatalities in Irish horseracing. Using race-event, horse-level and jockey-level data from 2010–2025, the project applies advanced statistical and machine learning models to identify key predictors of adverse outcomes. The aim is to generate evidence to support safety policy, prevention strategies and improved health monitoring in the racing industry.