Precision Medicine Startup FDNA Introduces Next-Generation Phenotyping (NGP)

Published on:

The founders of FDNA were previously involved with Face.com, the first facial recognition technology that performed as well as humans. Now, this technology gains insights that support the development of diagnostics. We spoke to CEO Dekel Gelbman about the future of healthcare.

How would you describe FDNA in a few words?

FDNA uses artificial intelligence to transform big data into actionable genomic intelligence to improve diagnostics and therapeutics in the rare disease space. With the world’s largest network of clinicians, labs, and researchers specializing in genetics, we are creating the most comprehensive genomic database to support our mission.

What inspired you to create the startup? How did it all start?

The founders of FDNA were the inventors of the first facial recognition technology that performed as well as humans, deployed in Face.com and sold to Facebook. In searching for a meaningful use of facial recognition technology, the founders learned about rare diseases, where there is an enormous need for AI to help recognize symptoms, features, and genetics of the more than 7,000 rare diseases. That moment defined FDNA.

In what ways are you testing your technology?

We continuously test the efficacy of our technology and spend significant resources researching links between a patient’s clinical signs (phenotypes) and their DNA (genotype), which helps us create innovative approaches to gain insights that support the development of diagnostics and therapeutics.

Your app Face2Gene delivers genetic evaluations based solely on uploaded photos. What can you tell us about the technology behind this process and where do you see it develop?

Face2Gene uses a deep learning neural network pipeline to analyze facial photos. This is much more advanced than just taking measurements of points on the face. Our innovative approach distinguishes between fine grain details of facial photo to identify patterns linked to specific variations in the human genome. The system learns from diagnosed cases using crowdsourced data from a global user base. This includes de-identified photo analysis information and clinically annotated features from tens-of-thousands of patients with known genetic syndromes.

The more the system learns from the data, the better it becomes at recognizing disease-related features and at suggesting relevant syndromes, clinical features, and genes for a clinician to consider. Our analysis also integrates with variant interpretation systems that use our data to improve laboratory analysis. To date, our system already covers thousands of genetic disorders.

You’re changing the game for rare disease patients – still, the competition appears to be fierce. What makes FDNA stand out of the crowd? What is the key differentiator?

Our key differentiator is that we focus on phenotyping (i.e., capturing the clinical signs caused by genetic variations). Genetic testing (especially whole genome sequencing and whole exome sequencing) is becoming central to virtually every vertical of medicine; yet, there are virtually no technologies available to capture and understand the physical expression of those genetics (phenotyping), making the interpretation of these tests very challenging and less effective. Our core product, Face2Gene, uses next-generation phenotyping (NGP) technologies, capturing a variety of biometric data (first and foremost facial photos) from patients and linking them to possible genetic variations that cause disease.

We’ve recently featured you in our Breakdown on Startup Driven Innovation in healthcare. What other cutting-edge technologies do you think the sector will experience in the coming years?

AI & big data are key to the future of healthcare. It allows us to gain a better and more granular understanding of medical problems and then address these problems with innovative solutions. Technologies that will increase the available data for analysis, such as genome sequencing, as well as sensors that collect biometric data from patients, will make a huge impact on understanding.

In addition, technologies that can use the insights from the analysis of the data, such as gene editing, will become essential for the future of medicine. This is the reason we have invested so heavily in being the leader of translating biometric data — such as facial analysis — into actionable genomic insights.

What is the biggest challenge that the company has faced?

There are more than 7,000 rare diseases — each of which has very few diagnosed patients around the world. Developing a technology that is trained to recognize the phenotypes of these rare syndromes required us to overcome the hurdle of gathering enough data on each disease. We overcame this challenge by getting tens-of-thousands of case files through our clinical partners globally, paired with our amazing team of engineers who used transfer learning approaches to first train our system on the general population before refining the system using the rare disease data we had obtained.

The second challenge we faced was privacy concerns since we were using facial images. We solved this by developing an approach that de-identifies photos in a way that prevents identifiable images from being viewed by humans, but still allow our machine learning pipeline to detect subtle patterns and link them to genetic variations.

What’s next for FDNA?

Bringing next-generation phenotyping (NGP) to the world, translating all relevant biometric data into actionable health insights. The majority of clinical geneticists around the world are already using our technology. It is also used by a large body of researchers, drug developers and genetic testing labs working to find answers and treatments for hundreds-of-millions of patients globally. In the future, every person’s genome will serve as their medical record. Our technology will help patients and their caregivers understand their health better and come up with a personalized care plan — this is our role in the future of precision medicine.

If there is one thing you could wish for in improving the European startup ecosystem – what would it be?

Working in Europe has provided access to a broad diversity of experiences and values that have shaped our product and goals. Each country houses specific passions, clinical approaches and regulatory environments that have taught us how to build a product that is respected cross-culturally and is value-added globally. Because there is no central organization of these perspectives, we had to invest heavily in each market to gain these benefits.

What’s one piece of advice you can give to fellow founders for their startup?

Do just one thing, and do it great.

 

__________

Sharing is caring!