The Danish Multigeneration Register

The Danish Multigeneration Register will use artificial intelligence to transcribe old handwritten church records and build a multigenerational register of family relationships for Danes born as far back as 1920. The register will strengthen research in Denmark and provide new insight into how hereditary and family relationships affect the trajectory of health and social life.

Denmark and the other Nordic countries are internationally recognized for the quality and completeness of their health registers, biobanks and archives, which have provided valuable insight into factors affecting health and disease. What is particularly unique internationally is the ability to link data across registries via the civil registration number in Denmark’s Civil Registration System. Unfortunately, health and social mobility can only be studied today across a few generations because the Civil Registration System, which was started in 1968, only contains family relationships for people born after 1960.

Based on the new epoch-making multigenerational registry, the Danish National Archives and its partners will create opportunities for research across generations through a much longer time frame, so that data can be linked from the years before the Civil Registration System was created. The multigenerational registry will be able to identify family relationships between all Danes born as far back as 1920.

Family relationships may explain many social and health problems, so giving researchers the opportunity to study phenomena over 3–5 generations will produce much greater insight into how hereditary factors and family relationships affect the trajectory of health and social life. Researchers will be able to use this knowledge to develop better treatments, including developing personalised medicine, and to prevent social and health problems.

Data science researchers will develop the algorithms that can decipher the handwritten church records. This task is difficult, because the handwritten church records from more than 2,000 parishes over almost 60 years represent a wide range of handwriting styles, most in cursive handwriting. The algorithms will be trained on a set of manually transcribed church records and combined with other public sources of information about names, dates and parish to build the registry.