By David Graff, AAU Communication & Public Affairs 

When a larger proportion of people in one area than in others are affected by a particular disease, there is reason to investigate the matter further, according to Professor Rasmus Waagepetersen in Department of Mathematical Sciences at Aalborg University.

"This is about investigating whether there is a correlation between the occurrence of the disease and the places where the patients live. A correlation that may explain whether the disease has a specific cause," he explains.

The discipline is called spatial epidemiology, and it can shed light on whether the occurrence of diseases is related to geographical factors. For example, Rasmus Waagepetersen has studied whether traffic pollution is a risk factor for pancreatic cancer.

From maps to microscopy

While the example of pancreatic cancer can be illustrated by plotting the occurrences of the disease as points on a map along with measurements of traffic pollution, Rasmus Waagepetersen and his colleagues zoomed in on the smallest components of the body in the department's latest project on spatial epidemiology. The focus here was analyses of skin samples from people with psoriasis.   

"These are also spatial point patterns, just locations at the microscopic level of tiny, tiny immune cells in the human tissue," says Rasmus Waagepetersen.

"And even though the scale is different, the method is basically the same: finding out how the points you are now analyzing are distributed, how and if they form clusters, what the distance is between the points and so on."

By using mathematical and statistical models to describe clustering and spatial patterns of immune cells in psoriasis data, the stages and development of the disease can be described more accurately, providing healthcare professionals with better, data-driven opportunities for earlier diagnosis and optimal treatment.  

Can predict development

The mathematical and statistical models used by researchers in spatial epidemiology also contain a rare possibility: They can predict developments to some extent.   

"When we observe spatial point processes on maps or in microscopy, we get snapshots of the situation at a given point in time. But you can also use some models to project how, for example, the spread of a disease in a population or the development of a disease in an individual patient will proceed," explains Rasmus Waagepetersen.

When the datasets become sufficiently large, it is possible to identify typical processes, including how situations at a given time are likely to unfold in the future. 

"But as we saw with COVID-19, for example, in practice there are often many unknowns. During COVID-19, we never knew all the individuals who were infected. And such dark figures challenge the possibilities for projection." 

Large amounts of data are a challenge

Another challenge for spatial epidemiological analyses is, paradoxically, the very large amount of data that researchers have access to today.

"You would think it’s a luxury to have access to very large amounts of data, for example from microscopy images of skin from psoriasis patients. But our models are typically developed to extract information from datasets with perhaps a few hundred points. When it comes to millions of points, it becomes, to say the least, complex," Rasmus Waagepetersen elaborates.

The development of supercomputers that can crunch the data and analyze even very large, complex datasets helps with the challenge, but nevertheless we need to develop our methods, Rasmus Waagepetersen believes:

"Most of the time, you want methods that are more manageable. For example, if you have a specific patient in need of treatment, there may not be time to wait for a supercomputer to be available."

Patterns in more than diseases

The wave of data that is sweeping over the healthcare sector by virtue of things like new microscopy technologies, is also reflected in many other areas. 

The result is that even in unexpected contexts, interesting patterns emerge where statistical and mathematical analysis can wring out useful information.

"Right now, I'm working on whether you can influence the development of urban areas by placing certain shops in certain places. This is based on point analyses of businesses in the Netherlands. For example, if you have a somewhat declining village, it’s good to know whether putting a large shopping centre there will bring about positive development, or whether it will be the beginning of all the small specialty shops dying out."

In the catalogue of spatial statistical research projects at the Department of Mathematical Sciences, you will also find, for example, analyses of piracy and biodiversity. And there is much more to come: 

"There is a lot of data that in one way or another is point patterns. So the potential is huge, and we aren’t exploiting it enough today. There is a lot out there that we could and perhaps should know that we don’t know yet," concludes Rasmus Waagepetersen.