Hello everyone, I’m late to the party here. On Saturday I stumbled into Tetiana Ivanova - How to become a Data Scientist in 6 months a hacker’s approach to career planning. Near the beginning of that presentation, Tetiana mentions that she was inspired by this guy called Jeremy Howard. The reference led me to the TED Talk which lead me to Enlitic. The following Monday I’m able to drive from my home in the redwoods here in Bonny Doon to San Francisco for the class. The Bay Area never ceases to amaze me! Thank you all for having me.
Two years ago I quit my job at a hedge fund to start Nourish Balance Thrive, a small functional medicine practice for athletes. I have two undergraduate degrees, one in electronics and one in computer science. I work with two medical doctors, one of whom is like me, a pro mountain biker, and the other is a biochemist and Ph.D. fellow. I employ a registered nurse and I work with my wife, a food scientist.
Over the past two years we’ve collected detailed health assessment questionnaires, blood chemistry, urinary organic acids, stool culturomics, and hormone data from around 800 athletes, many of whom are elite or even world-class. Our primary concern is fixing broken humans, second is performance. After a steep onramp, we’re now shifting away from exploration and into exploitation mode.
My burning question is: “can we automate the individualised diet and lifestyle medicine that we do?”
Second, I’m interesting in exploring our data for phenotypes. For example, we see lots of masters endurance athletes with iron overload. What other markers accompany that phenotype? Could something in a cheap test, e.g. heart rate variability or a complete blood count (CBC) be predictive?
Third, I’d like to build a learner capable of interpreting the scientific literature and use that knowledge to make diet and lifestyle recommendations. Do a Pubmed search for “diabetes or obesity”. Limit the result to the past 5 years. Last time I checked, that yielded 243k results. Who has time to comprehend all that? Only a machine.
I’m wondering if some of medical applications I’ve seen thus far are digging in the wrong place, so to speak. I’m less interested in drug discovery and early detection and more interested in creating systems that elicit the behaviour that prevent the problem in the first place. If you’re interested in what that behaviour might look like, I’d recommend my co-founder’s talk, A Systems Analysis to Insulin Resistance.
I’ve completed 60% of online training course called Creative Applications of Deep Learning with TensorFlow. It’s beautifully done and only $10 a month!
I’ve not been this excited about technology since the Internet, and I can’t wait to learn about your chosen applications.