Health Data Science and Informatics
Health Data Science and Informatics encompasses the re-use of clinical electronic health records for the real-world understanding of human diseases from the individual patient level to whole populations
Health data is routinely collected at all points of people’s lives - birth, death and many points in between. In the UK we have a unique health service where these electronic health records (EHRs) can be re-used for research to identify patterns of disease diagnoses, treatments and outcomes for the benefit of a healthier society.
This data takes many forms from columns of disease codes and rows of patient appointments to biosample test results, medical imaging and MRI/CT scans.
All this personal and sensitive information needs to be managed safely and securely. The Health Informatics Centre (HIC) is an accredited Scottish Safe Haven, also known as a Trusted Research Environment (TRE), where patient data for NHS Tayside, NHS Fife and NHS Forth Valley can be accessed within a computational architecture that adheres to the “Five Safes”. HIC has supported data-led research for over 20 years and more details of the range of activities undertaken can be found on the HIC website.
We are part of the Health Data Research UK (HDR UK) Alliance and host the £2.1m Alleviate Pain Data Hub as part of the Advanced Pain Discovery Platform which is one of nine HDR UK health data research hubs. A focus of Alleviate is to present research data using FAIR principles of Findable, Accessible, Interoperable and Re-usable such as standardising data via the Observational Medical Outcomes Partnership (OMOP) common data model and enabling international collaboration via the European Health Data and Evidence Network (EHDEN) data discovery initiative.
Machine learning (ML) and artificial intelligence (AI) are important fast emerging areas of research in Health Data Science. We are investigating its application to real world data in cardiovascular research on aging and early diagnosis of chronic conditions including complications of diabetes, neurodegenerative diseases such as dementia, heart failure and hypertension, various types of cancer, and its application for outbreak, infection and antimicrobial resistance (AMR) epidemiology and prediction. ML/AI models trained on sensitive data such as patient records pose challenges of additional disclosure risks to TREs which we are investigating.
Finally, we are working to ensure that our research findings are translated into clinical practice by developing clinical decision support systems within existing EHRs. We are working to improve the user experience of these clinical and patient-facing applications, and to evaluate the impact of these interventions on clinical processes and outcomes at a population level.
Principal Investigators
- Christian Cole
- Harry Hebert
- Charis Marwick
- Dan Morales
- Ben Parcell
- Huan Wang