With data collection becoming increasingly mainstream in business, the term “big data” has evolved into a broadly-applied, somewhat amorphous concept. Nonetheless, true applications of big data continue to drive tangible technological breakthroughs across industries. Healthcare is no exception. It’s no wonder, then, that half of the ten most innovative companies featured in Fast Company’s article “The 10 Most Innovative Companies In Health 2017” incorporate big data analytics and machine learning as a key component of their innovation. Ask a stranger on the street to tell you about today’s applications of big data and machine learning, and you’ll probably hear something about self-driving cars or video game analytics; but what GE Healthcare, Verily, Apple, AliveCor and others are realizing is that healthcare is as lucrative and as big of a machine learning playing field as any other industry– with perhaps an even larger upside.
It isn’t hard to see what all the fuss is about. Just looking at a few examples of machine learning in the healthcare space can get your heartrate up – Lumiata is using a predictive analytical tool to explore new insights and make predictions related to symptoms, diagnoses, procedures, and medications for patients. AiCure is leveraging facial recognition technology along with mobile capabilities to ascertain whether a patient is taking the right medication at the right time. More than one group is using machine learning to explore genomics and determine if early detection or even prediction of certain cancers is possible. Excited yet? And while the future of big data has never been brighter, getting to a place where it can generate return for an organization is easier said than done.
Machine learning and advanced analytics are the end-game, but there’s much to be done in the interim before companies can fully enjoy the fruits of this new paradigm. The first obstacle that we tend to see institutions face is escaping a pre-existing departmental, siloed approach to data. Missions, methodologies, and needs inevitably vary by department, so getting everyone on board with a big data lake-approach and operationally unified data ingestion process is a must. Once an aligned roadmap is established, organizations are almost always going to have to account for a disappointing user acceptance rate. Breaking down bad habits, ensuring users trust the new infrastructure, and getting cross-functional teams marching to the same beat takes a lot of collaboration and probably even a formal team to assess and ensure that new big data structures provide the core elements that departments previously relied on – in addition to calling out the new added benefits. Change is hard, the learning curve is steep, but broader, bigger, and better discoveries make it worthwhile.
Ultimately, big data planning and implementation proves to be as complex as it is to operationalize. In the rush to uncover and explore a realm that only machine learning and the non-human eye can deliver, it’s imperative not to forgo addressing the more mortal prerequisites to a successful big data platform.