At the core
Because of our diligence in collecting and recording information Denmark is the one country in the world with the most options provided by knowledge found in our databases. And because of these opportunities we are also well advanced in the development of mathematical models.
By Eline Mørch Jensen
What is cause and what is effect, what is the core when all the unnecessary layers are peeled off the ever growing amounts of complex data? To answer these questions researchers can´t ignore the use of mathematical models. Not if you ask statistician, PhD and associate professor at the Department of Biostatistics, Theis Lange, at least.
His field is causal inference, i.e. the analysis of causal relationships, and for Theis Lange the mathematical models are nothing short of a precondition for moving forward with the evidence-based research, which in turn can form the basis for functioning programs, for example in the health field:
- Since you can’t treat people like you can mice you must instead mathematically deduce, for example, the impact the choice of a physician has on the condition of the patient’s health. That is, measure the effect of the choices the doctor makes but without all the unnecessary layers that disrupt the overall picture, and which the models fortunately allow us to peel away, explains Theis Lange, adding:
- Naturally the measurements are based on reality in all its complexity, but we need to connect the mathematical apparatus, use the full statistical toolbox so to speak, so that we sometimes deal with very sophisticated models and at other times need simplifications instead - depending on what we are looking for.
According to Theis Lange, one way to peel away the noise is to use a measuring technique, Instrumental Variables Techniques, that is not directly involved in complex feedback mechanisms and among other things utilizes and draws inspiration from different models in nature, such as the frequency of lightning strikes in a given area at a given time.
He has, among other things, used this technique in a collaborative effort between the Department of Biostatistics and the National Research Centre for the Working Environment where a model was built to measure the effects of two different approaches at returning long-term sick-listed persons to the workplace. The method simply consisted of having the caseworkers suggest the sick persons one plan Monday to Tuesday and a second plan Thursday to Friday, allowing the researchers to read the pure performance without getting noise on the line in the form of questions about personal motivation, etc.
How is such a model developed in practical terms - where does the inspiration come from?
- Well, it 's hard to say where the good ideas come from… In the case of the long-term sick-listed persons I spent some time talking with people from the National Research Centre for the Work Environment about what they already knew, and especially what they felt needed detailed study and analysis. Generally I can say that you always learn something from talking to the people involved, so I spent a lot of time doing so ...
Because the different professionals can´t see the forest for the trees?
- Yes, or rather because we complement and stimulate each other when we cooperate across traditional disciplinary boundaries. Of course we regularly collaborate with different faculties; here at Biostatistics we primarily assist researchers in health science and to some extent also researchers in the fields of economy and work environment just as we also collaborate quite a bit with institutions such as Rigshospitalet, but there is a great - and growing - need to do more about it and to formalize the cooperation, says Theis Lange, adding:
- As an example there seems to be a parallel development on at least some points within the economical field and the medical profession respectively. Not because these groups are working with the same questions, but the way they do things, their tools, are more alike than you might think. The same may apply to the psychologists, but if I knew exactly what models they are working with there would be no need for research projects such as "Dynamical systems".
The problem is, according to Theis Lange, that the researchers are likely to waste energy by trying, within their own little narrow area, to reinvent the wheel over and over again, because the sharp demarcations between the fields limit the knowledge of the models used by others.
- So partly I expect a lot from the project’s interdisciplinary cooperation between the various parties involved here at the University since it is precisely in this area between practitioners and theorists that things really move forward, and I am sure that researchers at international level will benefit from this cooperation too.
What can Danish researchers add and contribute on an international level?
- I am convinced that this network will be able to address a great many medical, economic and social policy issues which will also be of great importance to researchers not directly involved in the project, both at home and abroad. Of course it is extremely difficult to predict exactly what issues, but I hope and expect that we will develop much stronger tools to handle many types of problems, says Theis Lange.
- Denmark is a pioneer in the development of models for the analysis of complex data, probably partly because of our unique record of social security numbers and collection of data about almost everything. The whole world comes to us to learn from the records we have, i.e. the fact that we can link between social security numbers and any other information we have on file.
- We can, for instance, compare information on income, education, children and housing to diseases that occur later in life. Take just such an example as to whether the use of a particular type of birth control pill increases the risk of blood clots. That is simply not possible to investigate outside of Scandinavia, but here it is possible because the pill is on prescription and because these are recorded and can be linked to our hospital records.
- Because of our diligence in collecting and recording information Denmark is actually the one country in the world where the knowledge stored in our databases provides the most options for research. And exactly because we've had these opportunities we are also quite advanced when it comes to developing mathematical models. We have been alone in this field somewhat, and that is probably one of the reasons why we have become so good at it ...
You do not see any ethical problems in the massive registration or a danger in parallel processing databases?
- There is no doubt that a dilemma exists here although I as a researcher obviously do not have access to any social security numbers when I retrieve information in the databases at Statistics Denmark, admits Theis Lange, adding:
- The challenge or dilemma lies in the very fact that this collection and registration even takes place. Once the data are collected, it would frankly be immoral not to exploit the opportunities which this gives us. This information does give us the chance to make people’s lives better.
About Theis Lange
Theis Lange holds and a MSc in Mathematical Economics and Econometrics from London School of Economics and in Mathematics-Economics from the University of Copenhagen.
He received his PhD in mathematical statistics (specialized in econometric) in 2008 from the Department of Mathematics, University of Copenhagen. He worked at the Department of Economics before joining the Department of Public Health, Section of Biostatistics in 2009. He has been head of this group since 2013. Associate Professor since 2011.