糖果派对

By Torben Skov Dyg Johansen, Assistant Professor, Department of Economics

Community health nurse visits for newborns are an important early intervention that has helped parents give their children the best possible start in life for nearly 100 years. With a health and care sector strained by labour shortages, there is a risk that these early interventions will be cut without fully considering the benefits they bring.

So how should we prioritise when resources are tight? How do we make sensible decisions about interventions such as community health nurse visits, where the benefits extend over long time horizons and can be difficult to measure?

It is difficult to put a clear value on how early support benefits children’s health and development. To measure the overall effects, we need to be able to track children over large parts of their lives. Although Denmark has high-quality register data, the data does not go back far enough. So we have to rely on archives and historical records.

This is where artificial intelligence (AI) comes in. New developments in AI make it possible to use such data to a greater extent than ever before. AI can now read handwritten documents more accurately than humans. This has led to a data revolution that allows us to collect information and study long-term effects in unprecedented ways.

Using this kind of AI, we at the University of Southern Denmark, in collaboration with the University of Copenhagen and Frederiksberg Hospital, have transcribed almost 100,000 community health nurse records covering all children born in Copenhagen between 1959 and 1967.

This data contains detailed information about the children’s development in the first year of their life, such as weight, breastfeeding, vaccinations and much more. In fact, the data includes over 10 million different measurements, which we have combined with Danish register data. This lets us track children from birth to around the age of 60, which gives us a better understanding than ever before of the long-term effects of factors like community health nurse visits and breastfeeding. This kind of data is invaluable, as it helps us understand which children and families benefit most from community health nurse visits, and which community health nurses are particularly effective with which newborns and their parents. Ultimately, it gives us a clearer picture of the overall impact of early interventions.

We can use this knowledge in two ways. First, we can design welfare initiatives to maximise their impact and give as many children as possible the best start in life. Second, we can learn more about how we should prioritise resources when they are scarce. E child and family is different, so it is important to maintain some degree of flexibility when designing interventions like community health nurse visits. Our research shows, for example, that extended community health nurse visits have enormous value for some – such as babies with low birth weight – whereas the benefits are more limited for others. This means that focusing on newborns who start life under the most challenging conditions could help reduce inequality in our society.

But there are not just differences among newborns and their parents – there are also differences among community health nurses. By examining differences between community health nurses and the children they visit, we have shown that there is potential for ensuring the best matches. Some community health nurses are particularly good at working with certain types of children and families, while others excel with different groups. This allows us to make even better use of community health nurse visits by thinking about how best to match community health nurses with newborns and their parents.

In the debate about how to prioritise healthcare resources, this kind of knowledge is crucial. First, to avoid cutting interventions that will pay off in the long term – especially early interventions, as research shows that the earlier they happen, the greater the benefit. Second, to better understand how to get the most out of early interventions. And finally, to ensure that when resource shortages in the healthcare system force us to make tough choices, we do so on the best possible basis.

By using AI to process historical data, we can combine modern register data with historical handwritten records, which can ultimately give us a more complete picture of the welfare state’s positive impact. This can help us design more precise, research-based interventions that are better suited to tackling our current and future challenges.

This feature was published in Jysk Fynske Mediers Erhverv+ on Thursday, 18 December 2024.