Big data refers to massive amounts of raw information that require new paradigms of storage, visualization, and analysis. With the ‘datafication’ of the world, these datasets are more available than ever, offering extraordinary opportunities to make predictions and discover new connections.

Attendees were taken beyond the hype of big data at Monday’s plenary session, where Viktor Mayer-Schönberger offered real examples of how big data analysis has revolutionized traditional analytical approaches. Big data is transforming hypothesis testing, where millions of ideas can be tested simultaneously rather than having to limit focus to a single question due to data sampling restrictions, Mayer-Schönberger explained. With “small” data, a researcher has to choose just what to focus on. With big data, she does not need to make the choice, and can grapple with everything, “zooming in and out at will,” he said. “Letting the data speak.”

For evidence-based medicine, algorithms can help make predictions even in the absence of an understanding of causation, and multiple correlations may be enough to derive useful meaning. “If correlations yield insights, then we should use them,” Mayer-Schönberger said. “We should go for causality if we can get it, but don’t put it on a pedestal as the be all and end all.”

Mayer-Schönberger also asserted that, in healthcare, we are over or under-treating all the time. Best practices are based on averages, where no particular treatment or therapy is really optimized for an individual. As our genetic makeup, metabolic state, and disease severity become better known and understood, personalized medicine will rely on big data. With more information from sensor technology, real-time measurements of basic chemistry analytes can lead to remarkable improvements in outcomes. Mayer-Schönberger described an example of real-time monitoring in preemies, where thousands of data points per second can predict a blood stream infection before it is detectable by traditional methods.

Other big-data leaps for healthcare include unusual corporate ventures. Mayer-Schönberger described how FICO, traditionally known as a credit score company, was able to use third party data sources to predict medication adherence. Incredibly, is now possible for a single entity to predict whether you will default on your mortgage, get a car loan, qualify for life insurance, and be compliant with your drug prescriptions.

Of course, big-data predictions are not perfect. Mayer-Schönberger described the Google Flu project, where flu outbreaks are predicted from keyword searches combining known outbreaks with epidemiological data from the Centers for Disease Control and Prevention. The Google Flu project has been criticized recently for apparent inaccuracies in the model. Society will certainly have to tackle difficult questions resulting from big data and predictive analytics. There are many fascinating—and perhaps somewhat frightening—implications to this new world of big data. One safe prediction is that policy makers globally will be at least a few steps behind.

“Big data will help us understand the world better and make better decisions,” Mayer-Schönberger said. “But also brings new challenges and dramatic dangers. We must remain its master. With a vital need to learn from the data, we need to allow for human creativity, and even defiance against the data. Data is a shadow of reality and therefore always incomplete. We must use humility and humanity.”