Predicting biomedical outcomes with big data
Social media “big data” can provide valuable perceptions about people’s behaviours, for example, their possibility of taking part in risk behaviours or contracting a virus. Although in its beginning, developing this research provides the potential of predicting health-related activities to promptly prepare for and respond to epidemics and public health emergencies. Making use of big data in the health sector can provide us with highly valuable insights about people’s health.
The field of big data science is a freshly evolving interdisciplinary field connecting researchers in areas as broad as computer science, public health, genetics, statistics, and the social sciences. Even though there is no absolute definition, big data is typically labelled as involving datasets categorized by complexity and enormity, rather than smaller scale data sets where memory-rich technologies are not required for preparing.
Predicting epidemics: Big data in the health sector
Big data science is significant because technologies, for example, smartphones, wearable gadgets, and portable diagnostic kits, have become gradually prevalent and affordable, giving large datasets accessible for merged analyses, including health datasets, genomics data, social media data and technology data. Social media users are becoming progressively comfortable sharing publicly any types of information, including individual stories and health data, providing information that can be extracted, categorized as behavioural and psychological data, and used for health research and analysis.
|Recommended for you
|How medicine is embracing social media?
|Engaging patients in healthcare
|Facebook’s project decodes signals from the brain’s speech center
Big data with social media can filter out indication for health emergencies and epidemics
Big data may contain not only relational and organized data, but also unstructured (e.g., text) data from assessments of social media conversations. For example, search engines and social media sites can be utilized to gather simple messages, searches, posts, and updates that deliver information about users. Social media platforms like Facebook, Google+ and Twitter, allows users to freely and easily connect with each other by sharing short messages, pictures, website links, and other multi-media messages.
Social media sites are turning into an essential part of big data research due to the high engagement of social media users (example over 2.7 billion likes per day on Facebook, 500 million tweets per day on Twitter). These datasets can be exhibited alongside other biomedical datasets and used to forecast biomedical consequences.
What does this mean for public health management?
Social media-based big data approaches are viable and can prompt to potentially revolutionary tools for public health administration and monitoring, they require
- the existence of interdisciplinary teams and methods, and
- the accessibility of large and regularly updated datasets.
An interdisciplinary team is desired to perform research using social media for biomedical big data analysis because this approach assimilates the fields of medicine, psychology, computer science, and business in addition to ability in the area of the first outcomes that will be evaluated. For example, to avail social media data, a collaborator with proficiency in Information Technology must be available to develop the structure for collecting and storing data.
To illustrate and label the free text from social media discussions (example, categorizing tweets or posts by whether they suggest an individual will involve in health-related behaviors), precise search terms and models (such as natural language processing) need to be extracted and advanced with the help of an expert.
For instance, an epidemiologist would be noteworthy if populations and disease statistical data were to be removed from the free text; and a geneticist or a basic scientist would be required to extract and label text associated with their fields.
To create methods via social media for the prediction of risk behaviors and disease, regular updates of both social media and biomedical data are wanted. Social media data can be queried, requested and can be received in real time; still, for biomedical data, there is often a delay in time between when individuals get an infection or sickness and the arrival of these data.
For example, a study on Google Trends had revealed that Google searches of flu symptoms could be utilized to anticipate outbreaks of influenza, the legitimacy of methodologies that use big data to foresee disease outcomes has recently been used.
Amid the Ebola crisis, UNICEF leveraged two existing dominant data platforms to notify its work:
U-Report– which is a framework that lets two-way real-time communication with young people via SMS in Liberia. For, e.g., youngsters required precise data about signs, side effects, and approaches to avert Ebola to battle the many rumors that were spreading within neighborhoods.
Edutrac, based on a similar innovation that powers U-Report, permitted to collect real-time data about the requirements in schools in Sierra Leone. It made sure that hygiene equipment had been transported to schools.
Amadeus and UNICEF have collaborated to facilitate the sharing of dynamic travel data to comprehend better patterns of the Zika Virus spread and possible epidemic zones.
IBM is additionally working together with the New York-based Cary Institute of Ecosystem Studies to gather and extract biological and ecological data to help devise systems that can figure out the carriers for the virus. IBM also runs the ‘OpenZika project’ which lets scientists in the Americas to screen millions of chemical compounds to identify candidates to fight the infection.
Social media data are rapidly impacting big data research and becoming one of the standard tools used in this developing field. Since these advancements provide rich biomedical data and individuals are openly sharing personal health information on social media, these innovations will keep on being observed and investigated for their potential in anticipating health-related situations and behaviors. Understanding the constraints of social media-based big data research (example, the legitimacy of data, missing data, observational data sample) and addressing these limitations will advance the value of this research in monitoring epidemics and health behaviors.
Image credit: www.istockphoto.com