Registry data and suicide prevention
Busenur Kizilaslan is a researcher and statistician at the National Centre for Suicide Research and Prevention (NSSF), University of Oslo. She holds an MSc and PhD in Statistics and has a research background in predictive modelling. Today, she applies statistical and machine-learning methods to investigate questions in mental health, including suicide and self-harm.
Her interest in healthcare research developed through earlier work at OsloMet and in the PharmaSafe research group at the University of Oslo, where she became involved in research using registry data.
Kizilaslan works on the EU-funded PERMANENS project, led in Norway by professor and director of NSSF Lars Mehlum, MD PHD, as part of an interdisciplinary collaboration involving researchers from Norway, Sweden, Ireland and Catalonia (Spain). The project explores how routinely collected registry data can support the assessment and management of people presenting with self-harm in emergency departments.
What can registry data tell us?
Registry data are valuable because they provide large-scale, longitudinal, real-world information that can help us understand suicidal behavior and risk factors, evaluate healthcare outcomes, and support evidence-based clinical and prevention strategies.
At NSSF, Kizilaslan works with Norwegian registry data and collaborates with researchers in partner countries. Her work involves curating registry data and applying statistical and machine-learning methods to investigate suicide and self-harm outcomes.
The project combines insights from national registries across participating countries, with each partner working with their own data to investigate suicide and self-harm outcomes. This approach also provides an opportunity to explore differences across registries and healthcare systems.
From models to clinical care
Developing the models is only one part of PERMANENS. The project also involves clinicians and people with lived experience of self-harm, bringing their perspectives into the development of the system.
The final system combines risk assessment with evidence-based clinical guidance and includes a training website to support clinicians in using the system appropriately. Together, these components form a clinical decision support system (CDSS) designed to support, rather than replace, clinicians’ professional judgement in clinical practice.
For Kizilaslan, this broader approach is important: making statistical information useful in clinical care requires not only good models, but also an understanding of how the information will be interpreted and used in practice.
Better decision support for suicide prevention
The value of machine-learning methods lies not only in their ability to work with large amounts of data, but also in how their results can be understood and used in practice. Transparency and interpretability are particularly important when models are used to support clinical assessment and care, and these principles are reflected in PERMANENS.
By bringing together registry data, clinical knowledge and professional judgement, PERMANENS aims to support more structured assessment and care for people presenting with self-harm. In this way, data-driven approaches may contribute to strengthening suicide prevention and improving the quality of care.