Boston startup FDNA, which makes a suite of apps called Face2Gene that use facial analysis, artificial intelligence and genomic insights in hopes of improving diagnoses and treatment of rare diseases, has launched a new tool for clinicians. Using de-identified data from their patients, clinicians can now share their findings and test and analyze patient cohorts with other clinicians around the world on Face2Gene’s Research application.
The new Research application builds on the Face2Gene app, which FDNA created out of a desire to create the world’s largest repository of rare disease big data. After snapping a smartphone photo of a patient the doctor suspects of, the app (which is only available to clinicians) analyzes the photo based on facial characteristics and compares the image against Face2Gene’s database, and within seconds, the clinician is offered a list of possible diseases along with the degree of probability for each one. While not a diagnostic tool, the Face2Gene app is intended to help clinicians as they go about testing for a specific condition.
The company – whose founders developed the facial recognition software purchased by Facebook – launched the Research application under their Year of Discovery initiative, wherein they hope to unite clinicians, labs and patients around the world to accelerate rare disease discoveries and understanding.
“When we started thinking about this, we made a very interesting discovery: it turns out rare diseases are not rare at all,” FDNA CEO and cofounder Dekel Gelbman told MobiHealthNews in an interview. “They make up a huge group of people around the world, and if you group them together, there are ways they can be evaluated.”
In the United States, about one out of 10 people (30 million total) are living with a rare disease, and 350 million people worldwide. On average, it takes about seven years from the time symptoms are present to get a diagnosis.
“Our mission is to save lives by reducing the diagnostic odyssey these patients go through,” said Gelbman. “The problem is the complexity of reaching a diagnosis, and so many physicians are unaware of the range of diseases because they are, by themselves, rare.”
But if there were a curated, aggregated database of all those different diseases, Gelbman said, they could then apply deep learning and artificial intelligence to bring the far-flung communities of people living with – and clinicians treating – rare diseases a little closer together. So FDNA spent the first three years of their existence as a company dedicated slowly to research, working with collaborators in clinical and academic settings to build up a trove of data.
“Since each syndrome is pretty rare, there isn’t a lot of value in creating a tool that would help the clinicians who already knew about it, so they secret was critical mass: getting enough collaborators and enough syndromes so we could start training a system to evaluate traits and classify into syndromes,” Gelbman said.
Operating in stealth mode for a while, FDNA was able to achieve just that. Since launching in 2014 with just a few dozen clinician beta users who were willing to share de-identified patient data, the Face2Gene database now features phenotypic and genotypic information associated with over 7,000 rare diseases, and the company reports over 65 percent of clinical geneticists around the world are using the app.
With the launch of the Research application, FDNA hopes to improve a number of enhancements to clinical studies, including the ability for researchers expand their data collections; reference Face2Gene’s data to support their hypothesis for publication; discover specific syndrome phenoytpes and their distribution within cohorts; leverage data from an increasingly expanding research community to compare gene cohorts to populations without rare diseases; and, overall, yield greater genomic insights.
From there, FDNA sees many implications for the technology. People with rare diseases often go without treatment or even research on potential therapies because pharmaceutical companies find little value in developing so-called orphan drugs. With FDNA’s large dataset of rare genetic cases, however, pharma could get a better idea of the sheer scope of people living with a condition and identify them for clinical trials. Treatments that do exist could also be started much earlier, if diagnosis is earlier.
“At the end of the day, it’s really about improving treatments for the people living with these diseases,” said Gelbman. “It’s a well known fact that recently, big data and AI will have dramatic effect on precision medicine and drug development. It’s even more prominent when you are talking about genetics. You can’t get more insights than with genetic big data.”