Biostatisticians use patient data to build mathematical models.

With the help of computer models, doctors can make more informed decisions

When a doctor tells you to go home to get sick, it doesn’t always feel good. You’re sick, why don’t you get medicine? After a few days, the doctor usually turns out to be right. Patience is a virtue. But although doctors do not make such a decision lightly, it is not always easy for them either. Every patient is different, so how do you know you’re making the right choice? Researchers from the field of biostatistics would like to help with this.

Biostatisticians analyze patient data and build mathematical models that map out the effects of different treatments, explains biostatistician Nan van Geloven of the Leiden University Medical Center (LUMC): “We can use the data to show which strategy seems most effective. for every patient, even for issues that are not suitable for traditional medical research.”

Traditional medical research often involves so-called randomized trials. In these trials, researchers divide a carefully selected group of people in two: half receive the drug, and the other half a placebo. “This kind of research provides the best evidence, but you can’t always do it. It is not always feasible or ethically justifiable to withhold treatment from people and see what happens,” says Van Geloven. “But we are collecting more and more data about patients in all shapes and sizes, and we can use that information.”

Building models

To extract information from the data, biostatisticians build mathematical models on the computer. For this, they often use an already existing basic model, which they adapt to the characteristics of the specific disease they are looking at. “In such a model you put factors such as age, gender, medical history and much more. Each disease has characteristics that are more important than the others. For example, older people may be more at risk of side effects from certain medications, or someone who moves more can recover from surgery more quickly. We take all that with us.”

If the researchers think their model can reasonably predict what will happen, they test it by inserting existing data and checking whether the results are correct. “Based on this data, you can fine-tune,” says Van Geloven. “It is important that you test your model a lot with different data sets, to minimize the chance of errors. And ultimately the model is useful, and you can enter the data from new patients to predict the effect of a certain treatment strategy.”

Wait and see attitude

One of the practical outcomes of this type of research is a tool within oncology. There, doctors can already use a model that indicates how different treatments affect the survival chances of a patient with breast cancer. Models for other diseases, such as infectious diseases, are still under development. What many of these models still lack, however, is the option to wait, according to Van Geloven. “Within oncology, for example, rapid treatment is usually better, but there are also many diseases where it can pay off not to start treating right away.”

For example, the biostatistician himself worked on a model for couples who want to become pregnant and are considering insemination treatment. “Insemination is quite a drastic process, both mentally and physically, so you don’t want to let women undergo this unnecessarily. That is why we analyzed data from 1800 women and looked at the effect of delaying treatment.” The model showed that a wait-and-see attitude could pay off: “A quarter of the women had become pregnant naturally after a year of waiting.”

Other studies show a similar picture. For example, sixty percent of the children with ear infections recovered after a day even without antibiotics, and some of the employees with back problems actually benefited little from the large reintegration projects of the UWV – even without these programs they managed it. to return to work in good health.

In the treatment room

The question, of course, is whether these kinds of results are actually used by doctors. According to Anna Roukens, internist-infectiologist at the LUMC, doctors generally take such new findings into account: “Most doctors keep a close eye on all developments in their field, and also take these into account in their decision.”

However, it can sometimes take a while before such a new discovery is incorporated into the official guidelines. “Within the medical world there are many guidelines, general frameworks that doctors adhere to, for example when to prescribe which medicine. Those guidelines are drawn up by a committee, and they should just be aware of these kinds of developments.”

But even with official guidelines prescribing patience, Roukens expects doctors not to be so quick to wait. “You want to be careful and make sure you don’t miss anything, and the patient expects something from you too. It is sometimes very difficult to say: just wait a little longer.” She sees many examples of this in her own field. “When someone with a respiratory infection ends up in the emergency room, we often immediately give antibiotics for five different bacteria, even if we don’t see any evidence for this at first. It would be better if we could treat more specifically on the basis of the symptoms or new imaging techniques. Because once you give antibiotics, it is also difficult to stop that treatment if you can’t find any bacteria – of course you don’t want to run the risk of missing something and wrongly stopping the medicine.”


Less treatment can also save costs. Not only for health insurers, but also for the patient.

Van Geloven also sees many forms of overtreatment: “Research has shown that we use quite a few treatments that have never really been proven to work. This doesn’t mean it’s bad for you, but you could have waited.” And that is why it is important that biostatistical research can support doctors. “They sometimes have to have difficult conversations, so it’s nice if they are supported by scientific evidence.” Roukens agrees: “You are looking for confirmation that you are making the right choices. For this you also use your medical intuition, the so-called clinical view, but this is supplemented by developments in science.”

In addition to doctors, health insurers also benefit from the research. After all, less treatment means less costs. Van Geloven certainly sees opportunities for health insurers to benefit from her models, but she disproves the idea that her research is driven by financial incentives: “Of course, our findings may ultimately yield cost savings, or influence what they do and do not reimburse. But the biggest benefits are always for the patient, because most medical treatments are really no fun. So it is very nice if we know better how to treat each patient. And if you no longer have to treat someone at all, that is of course the biggest gain.”

So if it is up to Van Geloven, biostatisticians will continue to do a lot of research in the coming years into when treatments are started. Although that is sometimes difficult. “You do need to have good data. For example, we want to test the model for insemination with larger data sets, to make sure it is correct. But they must be available. Fortunately, we are collecting more and more data, so that will eventually work out.”

Source: Kennislink by

*The article has been translated based on the content of Kennislink by If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!

*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.

*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!