The FDNA start-up, founded by Moti Shniberg, began developing medical software in 2014 called Face2Gene. This application is based on an automatic learning model (or machine learning) and on facial recognition in order to diagnose certain rare genetic diseases from the facial features of a child. In seven years, it has greatly improved.
A new machine learning algorithm to improve the Face2Gene software
It is thanks to an observation made by Moti Shniberg that this application was born: he noticed that many doctors used the facial features of children to detect a rare disease, a good number of these pathologies leave certain clues on the children’s face. This is particularly the case with Teacher-Collins syndrome, one of the symptoms of which is facial dysmorphism.
Facial recognition: Clearview AI technology used in Ukraine
The Face2Gene application has recently been modified using a new algorithm titled GestaltMatcherthe operation of which was explained in the review Nature Genetics. Thanks to this addition, the software is able to detect more than 1,000 diseases against “only” 300 previously. To achieve this, it identifies the signs and symptoms left by the disease on a child’s face.
To successfully train this brand new algorithm, scientists from FDNA, but also from several research organizations such as the University of Bonn, used 17,560 photographs of patients with characteristics linked to 1,115 rare diseases. According to its creators, GestaltMatcher is a much more efficient model than the one used in the past in Face2Gene.
This is what the Face2Gene application developed by the start-up FDNA looks like. Image: FDAA.
It is now able to detect a rare disease with fewer patients labeled by the application before: a real breakthrough. ” Similarities with patients who have not yet been diagnosed are taken into account by the model, which makes it possible to add combinations of features that have not yet been described “, specify the researchers.
FDNA helps doctors quickly diagnose rare disease with facial recognition
According to the start-up, almost 70% of geneticists in nearly 2,000 medical institutions around the world use Face2Gene in their research and in establishing diagnoses. Rare diseases affect nearly 300 million worldwide according to the National Institute of Health and Medical Research (INSERM) : “ This is an important contribution for very rare diseases, where only a few patients have been reported worldwide “, explain the researchers involved in this project.
The objective of the application is to detect a maximum of rare diseases so that they are diagnosed as soon as possible, at an early stage. Thus, doctors could offer as quickly as possible a treatment to treat children affected by these diseases. FDNA promises that all data collected through its application will be anonymized and stored in a ” private and secure cloud-based clinical warehouse “.
Facial recognition is increasingly used by many companies around the world. Until now, this technology has regularly been at the heart of controversy, in particular over its role in monitoring or even controlling populations, such as the Uyghurs in China. As often and as FDNA shows, the controversies surrounding new technologies are in reality linked to the uses made of them.