1. Building clinical trial machine learning software in R.
   —
   Many people think you can't use R to build production software, particularly when that software uses adaptive machine learning to process millions of data points, and those data are from patients on live pharmaceutical clinical trials studying new drugs for FDA approval. I'll explain the machine learning methods and R programming approach to solve this problem.

Bio: Mike Elashoff is Head of Data Science at Project Ronin, a start-up using personalized predictive models to improve treatment outcomes for cancer patients. He was previously a statistical reviewer at the FDA, and co-founded a software company that developed novel machine learning / AI methods for clinical trials (acquired by Medidata Solutions in 2014). He received a PhD in Biostatistics from Harvard University.

2. Combining similar data from different sources
—
The dirty little secret of data science is that we often spend much more time preparing the data for analysis than doing the analysis itself, and may even ignore data because it is too time consuming to add to what we have. Often we can find multiple sources with complementary information that need to be standardized to be used together (e.g. air pollution data reported by each state agency). I will share the packages, workflow, and philosophical insights that allowed us to standardize clinical trial data in an efficient, reproducible, and scalable way, and will hopefully help you create more complete datasets for your projects.

Bio: Clara Oromendia is a Biostatistician at Project Ronin, a start-up using personalized predictive models to improve treatment outcomes for cancer patients. She believes it is our duty as data scientists to add context to data and transform numbers into actionable conclusions in a scientifically sound way. After stints in Minnesota, Seattle, and NYC she is excited to have recently moved to SF and start enjoying the rich restaurant scene here.