In our market ascent, Nutonian has noticed that many of our most “cutting-edge” customers are biotech and healthcare companies. From drug demand forecasting to clinical trials analysis, Eureqa has been a critical cog in helping companies go from raw data to pattern detection, pinpoint forecasts, and root cause discoveries.
On the heels of this success, we posted up at booth 136 at Bio-IT World, a massive showcase of technologies enabling the biomedical, pharmaceutical and healthcare industries. A few lessons learned from the realms of artificial intelligence and data science…
There’s a problem with the status quo
…And it centers around analytics. Almost every attendee we talked to, from the analysts to the department heads, had massive amounts of data and deep domain expertise about their problem. But there was a glaring gap in their ability to extract equally deep understanding about what their data meant. They were like parents who were preeeeetty sure their kids were sneaking out at night but had no proof of anything – and they were setting up traps (manual, time-intensive statistical analyses) that were taking months to confirm or deny their allegations. It’s particularly painful to have this discussion with scientists who are spinning their wheels on problems that can immediately improve lives. If drug effectiveness for a particular disease can be accurately forecasted or optimized early in the R&D process, there’s a tangible value to the vendor and their patients.
R is still the modeling tool of choice
Based on an entirely unscientific survey of people who stopped by to chat, a majority of them are using R for their data science needs. R is, of course, not a biotech-specific tool, and their users face a grueling time-to-answer cycle, and at times, questionable model accuracy. Before a 5-minute Eureqa conversation, they had no idea how unhealthy and behind the times they really were. Shameless plug: Eureqa, with its academic roots out of the artificial intelligence lab at Cornell, was initially invented to accelerate scientific research intiatives and automatically unravel how scientific systems behave. It has since become popular in the private sector, unlocking the “rules of business.”
Big data is still a buzzword
Some of the smartest people in the world, with more of the right data than many other people in the world, are yet to incorporate much analytic technology into their processes. A common question we got was, “Are there data discovery techniques we can use to help us be more targeted in our research?”
Many times, the most overwhelming part of data science is simply figuring out what data is relevant. Multiple prominent universities and corporations were stymied by feature creation. They had so many thousands of variables and inputs in play during their research, that even the idea of analytical modeling was intimidating and unrealistic, and they were working off of trial and error hunches. Another shameless plug: Eureqa automates the entire process of feature engineering. Feed it all of your data at once, relevant or irrelevant, and it will tell you within minutes what matters and what doesn’t.
We’ll keep the community posted as interesting new use cases from the show materialize into Nutonian customers.