Artificial intelligence can be used to effectively diagnose rare diseases

Researchers at the University of Bonn have shown that using portraits in combination with genetic data and patient data improves diagnosis. Each year, about half a million children worldwide are born with a rare hereditary disease. Getting a definitive diagnosis can be difficult and time consuming. Scientists from the University of Bonn and Charity – Universitätsmedizin Berlin have shown that artificial intelligence can be used to diagnose rare diseases more efficiently and reliably.

A neural network automatically combines portraits with genetic data and patient data. The results are now presented in the journal "Genetics in Medicine". Many patients with rare diseases go through long trials and tribulations until they are properly diagnosed. "This translates into a precious waste of time, which is actually needed for early treatment to prevent progressive damage," says Prof. Dr. med. Dipl. Phys. Peter Krawitz of the Institute of Genomic Statistics and Bioinformatics of the University Hospital Bonn (UKB).

With an international team of researchers, he shows how artificial intelligence can be used to establish relatively fast and reliable diagnostics in facial badysis. The researchers used data from 679 patients with 105 different diseases caused by the modification of a single gene. These include, for example, mucopolysaccharidosis (MPS), which causes bone deformity, learning difficulties and growth retardation. Mabry's syndrome also causes intellectual disability. All of these diseases have in common that the facial features of the people affected have abnormalities. This is particularly characteristic, for example, of Kabuki syndrome, which recalls the composition of a traditional Japanese theater. The eyebrows are arched, the distance between the eyes is wide and the spaces between the eyelids are long.

The software used can automatically detect these features of a photo. With clinical symptoms of patients and genetic data, it is possible to calculate with great precision which disease is most likely to be involved. The artificial intelligence and digital health company FDNA has developed the DeepGestalt neural network, which researchers use as an artificial intelligence tool for their study.

PEDIA is a unique example of next-generation phenotyping technologies. The integration of an advanced framework for badyzing AI and face, such as DeepGestalt, in the workflow of the badysis of variants will give rise to a new paradigm for superior genetic testing. "

Dekel Gelbman, CEO of FDN

Researchers form the neural network with 30,000 images

Scientists have trained this computer program with about 30,000 portraits of people with rare syndromic diseases. "In combination with facial badysis, it is possible to filter out the decisive genetic factors and prioritize the genes," says Krawitz. "Data fusion in the neural network reduces the time of data badysis and leads to a higher diagnostic rate." The head of the UKB Institute for Genomic and Bioinformatics Statistics has been working with the FDNA for some time. "This is of great scientific interest to us and also allows us to find a cause in some unresolved cases," said Krawitz. Many patients are still looking for an explanation of their symptoms. The study is a team effort between computer science and medicine. This is also seen in the first joint publication of computer scientist Tzung-Chien Hsieh, doctoral student at Professor Krawitz's institute, and Dr. Martin Atta Mensah, doctor at the Institute of Medical Genetics and Human Genetics of Charity and fellowship of the Clinician Scientist Program at Charity and the Berlin Institute of Health (BIH). Professor Stefan Mundlos, director of the Institute of Medical Genetics and Human Genetics of Charity, also participated in the study, as well as more than 90 other scientists. "Patients want a quick and accurate diagnosis.Artificial intelligence helps doctors and scientists shorten the journey," said Dr. Christine Mundlos, Deputy Executive Director of the Alliance of Rare Disease Patients Chronic (ACHSE) eV "It also improves the quality of life of those affected to a certain extent." The results will be presented at an international conference

Scientists will present their study at the conference of the European Society of Human Genetics (ESHG) from 15 to 18 June in Gothenburg (Sweden). FDNA will also be present at the conference.