Big Data

Types and Uses of Big Data

The 2012 World Economic Forum estimates that sources such as blogs, e-mails, sensor data, search engine queries, and social networking traffic generate 2.5 quintillion bytes of data every day and will produce another 40 trillion gigabytes by 2020. The report lists several ways this information is shared by individuals and managed as big data in the public and private sectors. Information provided by individuals is often crowdsourced and is a by-product of other mobile transactions. The public sector maintains census figures, tax and financial records, and health system or facility data to improve services and increase efficiency. The private sector manages archives of financial transactions and spending information to predict purchasing behaviors and develop more detailed consumer profiles.

Because much of the data comes from easily accessible communication technologies—6 billion of the world's 7 billion people own a mobile phone—analysts can interpret the past, observe the present, and predict future patterns as individual indicators or population shifts. Many fields have taken advantage of this potential. Political scientists forecast voting behavior from social network traffic, criminologists intercept likely criminals from past crime data and, in health care and communication, researchers estimate virality both on the ground and over the Internet. Also, practitioners more accurately measure the efficacy of health messages, from news frames to fear appeals, without the limitations of small samples.

Big Data in Health Communication

A 2011 report by the McKinsey Global Institute identifies four types of data in the health domain. First, academic institutions and pharmaceutical companies maintain pharmaceutical research and development data, including records of clinical trials and high-throughput screening statistics. Second, health care providers own clinical data such as electronic medical records and medical images. Third, retail or apparel companies keep patient behavior and sentiment data such as purchase history and fitness information recorded by electronic exercise equipment. Finally, insurance companies and health care providers own claims and cost data, including utilization of care records and care cost estimates.

The report outlines a number of outcomes in health communication from mining and analyzing these data. First, big data permits more immediate tracking and earlier detection of epidemics. Using a Spatio Temporal Epidemiological Modeler, researchers at IBM simulate the spread of infectious disease and correlate it with other variables such as the reproductive behavior of animal carriers.

In a 2009 article in Nature, Jeremy Ginsberg and colleagues converted raw data from search engine queries for influenza symptoms into a real-time and region-specific tracker of the virus. The one-day delay was, at the time, up to two weeks ahead of the reports released by the U.S. Centers for Disease Control and Prevention.

...