AI could spot Genetic Disorders

Researchers at the University of Oxford have developed facial recognition software that scans facial images, analyzes subtle features, and compare the images to a database of faces of people with developmental disorders such as Down’s syndrome.

The computer detects similarities in facial features and makes a suggestion that you or a family member may have an undiagnosed genetic condition.

As detailed in the study published in eLife, the team at Oxford initially fed the software with more than a thousand photos of people who have been diagnosed with eight rare genetic disorders, namely: Angelma, Apert, Cornelia de Lange, Down, Fragile X, progeria, Treacher-Collins, and Williams-Bauren. An algorithm analyzed 36 specific features from each of the faces, including the shapes of the nose, eyes and lips, to determine which feature, or collection of features identify each condition.

Over time, the team added more pictures into the database, which now store 2754 faces, representing up to 90 disorders. The computer is continually learning, and is currently 30 times more sensitive in detecting the correct condition than by pure chance alone, according to an article in New Scientist. The more faces it analyzes, the smarter it gets. That may not be enough to make a diagnosis just yet, but its accuracy can help clinicians, the researchers claim.

The technology can scan ordinary photos, that includes old family photos and selfies, and look for similarities in the features of faces of those already entered in the database. The software accounts for differences in lighting, background, and image quality in its analysis of a photo. It then suggests — in order of likelihood — what genetic disorder an individual might have.

Clinical geneticists look at craniofacial features to help with their diagnoses because 30–40 percent of genetic conditions are thought to involve changes in the face and skull. Genetic disorders are estimated to affect one in 17 people. There are 7,000 inherited disorders, but these are rarely identified and treated because families do not have easy access to the few experts who can diagnose these conditions.

“A diagnosis of a rare genetic disorder can be a very important step. It can provide parents with some certainty and help with genetic counselling on risks for other children or how likely a condition is to be passed on,” Dr. Christoffer Nellåker of the MRC Functional Genomics Unit at the University of Oxford, lead researcher, said in a statement. “A diagnosis can also improve estimates of how the disease might progress, or show which symptoms are caused by the genetic disorder and which are caused by other clinical issues that can be treated.”

With this type of technology, patients and their families need not travel to hospitals and laboratories to get a formal diagnosis. Just a snapshot or selfie from a smartphone camera could be all that is needed to get it done.

“A doctor should in the future, anywhere in the world, be able to take a smartphone picture of a patient and run the computer analysis to quickly find out which genetic disorder the person might have,” Dr. Nellåker said in the statement. “This objective approach could help narrow the possible diagnoses, make comparisons easier and allow doctors to come to a conclusion with more certainty.”

Facial recognition software may be more known nowadays as a tool in fighting criminals, for photo-tagging in Facebook, or for identifying strangers for Google Glass users. Privacy issues notwithstanding, healthcare could benefit from facial recognition technology, especially if it means saving lives, or at least improving the quality of life of those with genetic disorders.

This technology could help thousands worldwide, but could be especially useful for citizens of regions like the Middle East, where there is a higher incidence of genetic disorders compared to most industrialised nations.