New AI medical program can spot rare diseases by looking at your face

BONN, Germany — When you’re sick, you can often see it in your face that you’re not feeling well. For rare diseases, it’s usually not that easy. However, researchers in Germany say artificial intelligence may change all that. A team from the University of Bonn say a new facial analysis program can actually detect the warning signs of rare diseases by examining the features of a person’s face.

“The goal is to detect such diseases at an early stage and initiate appropriate therapy as soon as possible,” says Prof. Dr. Peter Krawitz from the Institute for Genomic Statistics and Bioinformatics (IGSB) at the University Hospital Bonn in a university release.

The vast majority of rare diseases are genetic. Usually caused by hereditary mutations, each one of these rare genetic conditions usually impair the body in a variety of different ways. Importantly, though, most of these rare genetic diseases also result in specific facial features. For instance, the base of a person’s nose or cheeks may have a different shape if they carry a genetic disorder. Such facial changes, however, vary from disease to disease.

This newly developed AI looks at those subtle facial differences, analyzes them, calculates similarities, and then automatically connects them with the clinical symptoms and genetic data of specific patients.

“The face provides us with a starting point for diagnosis,” adds Tzung-Chien Hsieh of Krawitz’s team. “It is possible to calculate what the disease is with a high degree of accuracy.

GestaltMatcher can even spot ‘unknown diseases’

Named “GestaltMatcher,” this astounding AI system requires only a few patients to properly detect disease warning signs. This is a big advantage when it comes to rare diseases. Some diseases only have a few confirmed cases across the globe. Another advantage is the AI’s consideration of similarities among patients with no formal diagnosis yet, consequently accounting for combinations of disease characteristics doctors may not know about. These features mean that GestaltMatcher is capable of “recognizing” diseases that even the AI was unaware existed.

“This means we can now classify previously unknown diseases, search for other cases and provide clues as to the molecular basis,” Krawitz says.

The team used the photos of 17,560 patients during their project. Most of those images were provided by the digital health company FDNA, while another 5,000 images came from the Institute of Human Genetics at the University of Bonn and nine additional universities. FDNA assisted study authors with the development of the web service that facilitates the use of GestaltMatcher.

Study authors decided to focus on the most diverse disease patterns possible. Ultimately, they analyzed a total of 1,115 different rare diseases.

“This wide variation in appearance trained the AI so well that we can now diagnose with relative confidence even with only two patients as our baseline at best, if that’s possible,” Krawitz explains.

“We are very happy to finally have a phenotype analysis solution for the ultra-rare cases, which can help clinicians solve challenging cases, and researchers to progress rare disease understanding,” adds Aviram Bar-Haim of FDNA Inc.

AI facial diagnoses coming soon to smartphones?

Krawitz notes that the use of GestaltMatcher in local medical offices isn’t very far off at all. German doctors can already take a photo of a patient’s face on their smartphones and then use AI to make differential diagnoses.

“GestaltMatcher helps the physician make an assessment and complements expert opinion,” the study author notes.

Study authors have also handed off their database to the non-profit Association for Genome Diagnostics (AGD).

“The GestaltMatcher Database (GMDB) will improve the comparability of algorithms and provide the basis for further development of artificial intelligence for rare diseases, including other medical image data such as X-rays or retinal images from ophthalmology,” Krawitz concludes.

The study is published in the journal Nature Genetics.

Leave a Reply

Your email address will not be published.