GestaltMatcher, a new AI programme can diagnose genetic disease with a single selfie; Read

When you're unwell, you can usually tell that you're not feeling well by looking at your face. It's usually not that simple when it comes to rare diseases. Artificial intelligence (AI), according to German experts, might change all of that. By analysing the traits of a person's face, a team from the University of Bonn claims that a new facial analysis tool may truly detect the warning signals of rare diseases.

In a university release, Professor Dr. Peter Krawitz of the Institute for Genomic Statistics and Bioinformatics (IGSB) at the University Hospital Bonn said, "The goal is to detect such diseases at an early stage and initiate appropriate therapy as soon as possible."

The bulk of uncommon diseases is caused by genetic mutations. Each of these rare genetic disorders is usually caused by hereditary mutations and affects the body in a number of ways. Most of these rare genetic illnesses, however, also result in distinct face traits. If a person has a genetic disease, the base of their nose or cheeks, for example, may have a distinctive shape. However, the appearance of such changes varies from sickness to disease.

This newly developed AI evaluates those small visual changes, calculates similarities, and then automatically connects them to individual patients' clinical symptoms and genetic data. Tzung-Chien Hsieh of Krawitz's team added that the face provides us with a starting point for diagnostics. With a high degree of accuracy, it is able to calculate what the sickness is, Hsieh stated.

This incredible AI system, named "GestaltMatcher," just needs a few patients to correctly recognise disease warning symptoms. When it comes to rare diseases, this is a huge benefit. There is only a handful of confirmed cases of several diseases around the world. Another benefit is that the AI takes into account commonalities among patients who have not yet received a formal diagnosis, therefore accounting for illness traits that doctors may not be aware of.

The qualities allow GestaltMatcher to recognise diseases that the AI was previously unaware of

These qualities allow GestaltMatcher to recognise diseases that the AI was previously unaware of. This means we can now classify previously undiscovered diseases, seek for other cases, and provide molecular foundation hints, explains Krawitz. During the experiment, the team used photographs of 17,560 patients. FDNA, a digital health startup, donated the majority of the photos, with another 5,000 coming from the University of Bonn's Institute of Human Genetics and nine other universities.

FDNA aided the study authors in the creation of a web service that makes GestaltMatcher easier to use. The researchers chose to focus on the widest range of illness patterns imaginable. In the end, they looked at 1,115 different rare disorders. This huge diversity in appearance educated the AI so well, Krawitz continues, that we can now identify with relative confidence even with only two patients as our baseline at best if that's conceivable.

"We're thrilled to have a phenotyp analysis option for ultra-rare cases," says Aviram Bar-Haim of FDNA Inc. "It will assist clinicians solve difficult cases and researchers advance rare illness understanding."

According to Krawitz, the deployment of GestaltMatcher in local medical offices isn't far off. German doctors may already use their smartphones to take a photo of a patient's face and then use AI to make differential diagnoses. The study author writes, GestaltMatcher assists the physician in making an assessment and complements expert judgement. The authors of the study also donated their database to the Association for Genome Diagnostics (AGD), a non-profit organisation. The findings were published in Nature Genetics.

Image: Unsplash