A paper published in the American Journal of Medical Genetics in March demonstrated that it was possible to use facial recognition software to identify people with 22q11.2 deletion syndrome across diverse populations. Now, a Boston-based company called FDNA are hoping to use similar technology to identify a range of rare diseases with nothing but a picture on a smartphone. 

Face2Gene is an innovative smartphone application which allows users to take a photo of a patient, which can then be compared against a database of facial features to identify what condition they may have. Because of a combination of poor understanding and delays in complex genomic data interpretation, obtaining a diagnosis in rare disease cases can take years and involves significant medical costs. Face2Gene has the potential to provide a more inexpensive diagnosis in far less time.

“The idea was born from a previous facial recognition technology built for social networking purposes and acquired by Facebook,” said Dekel Gelbman, CEO of FDNA, when speaking with FLG. “We sought to find a medical application that would have a significant and positive impact on people’s lives. Since rare diseases affect almost 1:10 people globally, many of whom are children who go undiagnosed for long periods of time, we realized that with these technologies we could give hope to hundreds of millions of patients around the world and make cutting edge advancements in diagnostics and therapeutics for rare diseases.”

Currently the app draws on genetic information on more than 2,000 rare diseases, collected from across 129 countries. Over time, FDNA hope to expand the data pool the app offers, further increasing the breadth of diagnosis tools available to consumers.

Because of the personal nature of the photographs, complete with sensitive genetic information of the patients, there have been some concerns with the security of the app when storing these images. However, FDNA have assured customers that they have taken the necessary precautions to protect patient information and identities.

“Since facial photos are an important part of our analysis and by definition are considered Protected Health Information (PHI), we have devoted a lot of time and resources to ensure our system’s compliant with the most stringent privacy laws and security standards globally,” Gelbman told FLG.

“To overcome this challenge, we have built a system that stores all PHI in an encrypted and secure disk volume for each user (the Vault). Through a sophisticated set of algorithms, we are able to de-identify the patient information, by converting each photo into a mathematical description of the face, which cannot be reverse engineered,” he explained. “This information is fed into a deep learning system that continues to teach our system how to identify disease-specific facial patterns and correlate them with phenotypic traits and genetic variants associated with these diseases.”

Children with rare diseases take an average of 7 years to get a clear diagnosis, during which time their health may be deteriorating. Establishing a system that can improve this time frame is not only significant for the patients, but it can have knock-on effects for drug developers working on therapies for the disease in question. One of the biggest problems faced by these developers is finding humans to trial their drugs on, partially due to the number of undiagnosed patients.

“These technologies have already demonstrated an ability to improve the outcome of genome sequencing in research and clinical utilities targeted at rare diseases,” Gelbman said. “Therefore the future opportunities are endless and we foresee this technology being integrated tightly in the development of precision medicine that can apply to the entire population.”