Lead by Google Translate creator Franz Ocha, a team of data scientists are getting closer to cracking the DNA code and helping everyone live longer and healthier lives.
via Singularity Hub:
It’s a Translation Problem, Like Google Translate
HLI is creating an “integrated health record” for everyone entering its database. The data sets created will include the following:
- Genomic: The 3.2 billion nucleotides from your mother, and the 3.2 billion nucleotides from your father.
- Microbiome: The genome of the 100 trillion + microorganisms living in our bodies. There are 10 times as many microbial cells than human cells, and their effects on our bodies are enormous and massively understudied.
- Imaging/MRI: High resolution detailed imagery of our brain, organs and body.
- Metabolome: The 2,300 small molecule chemicals in your bloodstream.
- Physiological Health Data: All of the data we can collect on ourselves. Our vital signs, blood glucose levels, micro RNAs in the bloodstream, heart rate, VO2…
Translating between all of this data and your health outcome is, metaphorically, similar to how Google Translate works.
Google Translate (GT) uses a process called statistical machine translation, which means that GT generates translations based on patterns found in large amounts of written text.
Rather than attempt to teach the computer every rule of every language, this approach lets the computer discover the rules for themselves based on statistically significant patterns in the data.
Once it finds these patterns (patterns that are unlikely to occur by chance), it can use this “model” to translate similar text in the future.
With millions and millions of documents/websites/publications online that were already translated, and a crowd of 500 million users to correct and “teach” the algorithm, GT can quickly and accurately translate between 90 different languages.
Our challenge now is applying similar techniques to all of this genomic and integrated health records… and we found the perfect person to lead this effort: Franz Och — the man responsible for building Google Translate.
Meet Franz Och, HLI’s Chief Data Scientist
Franz is a renowned expert in machine learning and machine translation.
He spent 10 years at Google as a distinguished research scientist and the chief architect of Google Translate, literally building the system from the ground up.
Now, Franz is Human Longevity Inc.’s chief data scientist, responsible for developing new computational methods to translate between all of the human biological information.
… and he’s building one of the most impressive teams I’ve seen.
When you ask Franz why he’s so excited about HLI, his answer is twofold: the mission and the challenge.
Franz explains, “The big thing is the mission — the ability to affect humanity in a positive way. If you are a data scientist, why focus on making a better messaging app or better Internet advertising, when you could be advancing the understanding of disease to make sick people better and of aging to make people live longer, healthier lives?”
As far as the challenge, he goes on: “The big mission is to learn how to interpret the human genome — to be able to predict anything that can be predicted from the source code that runs us.”