Researchers are developing new algorithms to minimize diagnostic errors

Researchers at the University of Copenhagen have developed an algorithm that can identify patients who may have been misdiagnosed. With the help of a digital medical history, the algorithm can register a trajectory of the disease that is very different from the normal trajectory where there may be misdiagnosis. This algorithm was developed based on data from hundreds of thousands of COPD patients.

It doesn’t happen often. However, in rare cases, doctors can make mistakes and make the wrong diagnosis. Patients can get many illnesses at once, and it may be difficult to distinguish the symptoms of one illness from others, or it may be asymptomatic.

Misdiagnosis can lead to incorrect treatment or lack of treatment. Therefore, everyone in the healthcare system tries to minimize errors as much as possible.

Researchers at the University of Copenhagen are now developing algorithms that can do just that.

Our new algorithm can find patients with anomalous illness trajectories such that they may not actually suffer from the illness diagnosed. I hope it will be a support tool for doctors. “

Isabella Fleece Jorgensen, Postdoc, Novo Nodisk Foundation Protein Research Center

Algorithm reveals potential for lung cancer

Researchers have developed an algorithm based on the pathology trajectory of 284,000 patients with chronic obstructive pulmonary disease (COPD) from 1994 to 2015. Based on these data, we have devised about 69,000 typical disease trajectories.

“The National Patient Registry was able to map what could be called a typical illness trajectory, and if a patient appears with a very abnormal illness trajectory, it could simply be that the patient is suffering from another illness. It’s possible. Our tools can help detect this, “explains Søren Brunak, a professor at the Novo Nordisk Foundation Center for Protein Research.

For example, researchers found a small group of 2,185 COPD patients who died shortly after being diagnosed with COPD. According to researchers, it was a sign that something else might have been wrong, maybe something more serious.

“A closer look at the test values ​​in these patients revealed that they deviated from the normal values ​​in COPD patients. Instead, the values ​​were similar to those found in. Lung cancer patients.. Only 10% of these patients were diagnosed with lung cancer, but I’m reasonably convinced that most, if not all, of these patients actually had lung cancer, “explains Søren Brunak. ..

Data that brings immediate benefits

The algorithm has been validated through data from patients with COPD, but may be used for many other diseases. The principle is the same. The algorithm can use registry data to map typical illness trajectories to detect if some patients’ illness trajectories are too prominent and something might be wrong.

“Of course, our most important goal is for patients to get the value of money for their health care, and in the future this algorithm could be a support tool for doctors. When the algorithm maps a typical disease trajectory, it only takes 10 seconds to match one patient with all other patients, “says Søren Brunak.

He emphasizes that the algorithm needs to be further validated and tested in clinical trials before it can be implemented in a Danish hospital. But he wants it to be something that can be started right away.

“In Denmark, we often admire health registration because it contains valuable data for researchers. We may benefit other people in the form of better treatment in the future. Are using them in their studies, but this is actually an example of how your own health data can quickly benefit you, “says Søren Brunak.