DataBeat Recap: Integrating Data Science into Medical Research
By the time the DataBeat agenda moved to their Data Science in Action panel, we’d already heard about investments, education, the stock exchange, ethics, and the changing role of the CEO in a data world. All of these areas are seeing major shake ups due to the accessibility of data and the new tools and skills we have to analyze it, but none more so than healthcare, an area where data science is literally revolutionizing how we treat and prevent disease. We heard about this firsthand in a case study from Joel Dudley, Director of Biomedical Informatics for the Icahn School of Medicine at Mt. Sinai, and his colleague Pek Lum, the Chief Data Scientist and VP of Solutions at Ayasdi.
Pek Lum has a background in genetics; but she became a data scientist a few years back when genomes started to be sequenced. The potential for this technology is huge, and her new position as a data scientist gave her a great vantage point from which to watch the healthcare revolution.
Joel Dudley’s employer, Mt. Sinai, is leading the charge on this – the CEO placed a huge bet that data science will matter and ultimately change healthcare. In fact, no one has been more aggressive or committed to using data in healthcare. Mt. Sinai wants to use past data to continually improve treatment for the future, certainly a noble goal.
While they collect many types of data for many different reasons, Mt. Sinai has one particularly unique dataset they can draw on: a set of 25,000 patients with both clinical data (notes and medical history) and genomic data (information pulled at a genetic level).
Having both of these layers available puts researchers light years ahead of where they would be if they only had one type of data, since they can cross reference the two types of data and isolate ever smaller populations of people who may share traits that previously went unrecognized.
This type of research is data-driven: researchers don’t know what they’re looking for, but because they have multiple layers of data, they can compare and contrast patients with the same ‘conditions’ to see if any groupings become apparent. It’s a huge step forward in the clinical setting, since researchers don’t have to go into a study looking for a specific, previously identified issue (and thereby missing others, since it’s not their ‘focus’); instead, they can wait for a condition to show itself…and it just may be one that science has yet to uncover.
Dudley proposed a fascinating concept: will this be how we find diabetes Type III, Type IV, or Type V…? These labels may sound silly in a world where there are only two recognized types of diabetes, but data-driven research may soon uproot that. In this hypothetical situation, it may turn out that some of these newly isolated diabetic groups are more prone to comorbid heart attacks, whereas others are more prone to kidney failure. If the first group isn’t at all prone to kidney failure, but is still diabetic, then researchers very well may have found a new form of the disease, which will allow them to provide even more targeted, personalized treatment.
This is, of course, a hypothetical situation, but a fascinating one. Data science is allowing us to innovate healthcare in ways we would never have dreamed possible only a few short years ago – and data-driven research is just one example of a data science application that could have a very real impact on your own life.
Want more from DataBeat? Check out our Wednesday recap, or another in-depth post: