Rare diseases affect more people than one might think, but they are often difficult to diagnose. On average, it takes years for patients to receive a diagnosis. In this project, our researchers are working with the pharmaceutical company Chiesi to develop a mathematical support tool that will help doctors identify rare diseases more quickly and test for them in a targeted manner.
There are over 8,000 known rare diseases worldwide – no doctor will be familiar with all of them. This is where our research comes in: the tool we are developing together with Chiesi calculates the probability of various rare diseases based on the symptoms entered. In this way, we support medical professionals in narrowing down possible clinical pictures and initiating targeted tests.
What Defines Rare Diseases?
Rare diseases affect more people than one might think – but that is only one of their characteristics. Many have genetic causes and often manifest early in life. They frequently impose severe limitations, shorten life expectancy, and remain incurable to this day. Furthermore, effective therapies are lacking for many of these diseases. The goal of the project is to ensure that those affected receive the correct diagnosis and thus the appropriate treatment more quickly.
Our Methodological Approach: Bayesian Statistics and Monte Carlo Simulation
For the diagnostic tool, we combine Bayesian models with Monte Carlo algorithms. These mathematical methods make it possible to calculate reliable probabilities even with incomplete or uncertain information. This allows patient data, known symptom distributions, and prior medical knowledge to be intelligently linked. The more information is available, the more accurate the results are – meaning that the model can be continuously improved with new medical data.
A particular challenge lies in recording and modeling the symptoms: in the first phase of the project, the team focused on a smaller group of rare metabolic diseases to make the task manageable. For her bachelor’s thesis, student Chiara Freitag delved deeply into the symptoms and structured the available data from the perspective of those affected – that is, how symptoms are typically perceived and reported. This is because many patients do not have all the symptoms or are unaware of some of them because they can only be detected through special examinations. The stochastic model must therefore deal with incomplete and uncertain data, while at the same time integrating prior knowledge about symptom probabilities in order to make the most reliable statements possible about probable clinical pictures.
Cooperation Between Universities, Industry, and Research
The project was developed in close collaboration with Bielefeld University of Applied Sciences and Arts (HSBI). The bachelor’s thesis by Chiara Freitag, who is now a research assistant at Fraunhofer ITWM, laid the methodological foundation. The thesis was supervised by Prof. Jörg Horst from HSBI and Dr. Jan Hauth from Fraunhofer ITWM.
Based on this research, the pharmaceutical company Chiesi commissioned us to develop the diagnostic tool – the second phase of the project is now underway.
Benefits and Prospects: Effective Diagnostic Aid
Our approach does not replace a medical diagnosis, but it can significantly shorten the path to one. Doctors receive mathematically sound support in assessing rare clinical pictures, and patients benefit from earlier treatment and thus a better quality of life.
In the long term, the model can be expanded to include additional disease groups and adapted to new data sources. This is a good example of how applied mathematics at Fraunhofer ITWM enables concrete improvements in medical care.
You can find more information about the project here: www.itwm.fraunhofer.de/seltene-erkrankungen
