Psychiatric epidemiology with a life course perspective

This project utilizes the incredible opportunities in Danish Register data to do life course analysis since the registers date back a long time. For example, when looking at hospital the Central Psychiatric Register dates back to 1969 and later in 1995 we have information on other hospital data. These data are linkable to registers with information on social factors on an individual level, hence giving us plenty of opportunities for life course research.

In the first part of the project, we are using the data on hospital contacts together with other predictors such as level of education and income to predict dementia in old age. With prediction models the aim is to develop a model that can predict an individual’s risk of an event with high certainty. We are developing and using classic prediction models and language processing models, with the latter being more able to take the complex dynamics of disease into account.

The aim is to compare the two modelling approaches to see if the models taking the timing and ordering of the diagnosis into account will be more accurate than the classic prediction models with categorical predictors. Our aim is to build on the knowledge created and work made by Savcisens et al. (2023) in the life2vec work predicting mortality and personality and the previous work by Jørgensen et al. (2023) showing how often patients in contact with the psychiatric hospitals get a new psychiatric diagnosis.

The next few papers in this project are still in the early stages of development.

The project is run by PhD student Mathilde MB Sloth supervised by Samir Bhatt, Terese SH Jørgensen and Alexandros Katsiferis. Below you can see a full list of researchers involved in the project.