Big data is big news in many fields, including healthcare. Big data solutions analyze vast quantities of aggregated data from varied sources, providing more accurate analytics much more quickly than previous tools. An article in the Journal of AHIMA claims that big data can help healthcare providers to “personalize care, engage patients, reduce variability and costs, and improve quality.” Big data helps to personalize care by comparing large cohorts of patients matched on multiple characteristics, generating information about health factors and outcomes at a previously impossible level of granularity. Analyzing large volumes of clinical data from very specific populations in this way enables healthcare providers to predict more accurately the best treatments for each patient. Big data can also contribute to improved healthcare quality control and system planning, for example, by identifying settings where over-prescription is causing adverse health outcomes, or comparing patients’ trajectories through the healthcare system under different healthcare models.
Discussions of big data in healthcare often centre on leveraging electronic health records as a source of data. In fact, huge volumes of health data suitable for big data analytics already exist in the form of hospital records, clinicians’ notes and health insurance data, as well as consumer health applications.
Consumer health applications
Mobile health devices are an increasingly common type of consumer health applications used for day-to-day monitoring of basic health indicators. A variety of devices are used by individuals with chronic health conditions to monitor heart rate, blood glucose levels, blood pressure and other indicators. In some pilot programs, mobile devices are being programmed to upload results to online portals accessible to individuals’ healthcare providers. The information generated by mobile health devices helps healthcare providers to monitor chronic health conditions and engage patients in managing their health, and can enhance communication between patients and healthcare providers.
Another form of consumer health applications are health and wellness applications, which are used by a much broader population. Digital technology is one of the main sources for health information for many people, who use tools such as exercise and diet tracking applications and online health and wellness quizzes. These applications cover a wide range of common health issues such as nutrition, fitness, pregnancy, child development, substance use and mental health. Ideally, well-constructed applications from reputable sources can both help individuals to manage their health and collect data for public health research. By providing online information, questionnaires and applications, healthcare organizations could potentially provide patients with accurate health information, promote healthy behaviours and gather public health data from a broadly representative population.
Investing in big data solutions
On an aggregate level, data from consumer health applications opens up a number of analytical possibilities. Most health systems research to date is based on healthcare claims data and limited-scale program evaluation studies. Information from consumer health applications provides a more detailed, more frequently updated, and broader source of data on public health.
Big data analytics are a significant investment, and it is necessary to consider where this investment would be most effective. I would argue that data from consumer health applications is an ideal source for big data analytics. Consumer health data can enable organizations fully to utilize many of the advantages of big data. Fernandes, O’Connor and Weaver describe the key characteristics of big data as volume, variety, velocity, veracity, validity and volatility.
Big data solutions can process very large quantities of data in near real time. Consumer health applications provide large quantities of frequently updated data. Analyzing large volumes of data, particularly longitudinal data pertaining to the same individuals, using big data tools makes it possible to detect patterns with greater sensitivity and specificity than previous analytics tools.
Variety refers to multiple sources and types of data, including both structured (e.g., research results) and unstructured (e.g., clinicians’ notes) data. Mobile health devices provide information mostly pertaining to patients with the most complex health needs. Health and wellness applications, on the other hand, provide data pertaining to a very broad population. Consumer health applications offer multiple forms of data pertaining to specific health conditions as well as the health of the general population.
Big data solutions collect and analyze data nearly instantaneously from sources including mobile devices, personal computers and dedicated hardware sensors. These tools can capture the massive and continuous flow of consumer health data, and provide up-to-date analytics to inform healthcare decisions. For example, consumer health applications could be used to gather outcome data on new treatments or healthcare models much more frequently and on a broader scale than current evaluation tools. Particularly with regard to new healthcare models, current analytics based on sources such as healthcare claims data can take years to evaluate the success of a new model; big data analytics would make timely, evidence-based decision-making possible.
Most healthcare data from clinic and hospital records is plagued with errors, as technicians entering data frequently attach information to the wrong person’s record or copy information incorrectly. Consumer health applications are largely automated and have simple interfaces, reducing the potential for human error, and thus contributing more accurate data.
Consumer health data, especially data from mobile health devices, has a relatively short life span: it is usually used to provide individuals with immediate feedback on their health and any actions they need to take to manage a condition. Big data solutions are able to process large amounts of data quickly without needing to retain it indefinitely. On the other hand, big data solutions can also be used to follow individuals’ health over time much more efficiently than current longitudinal research methods.
Volatility refers to how long data should be stored. Much of the data from consumer health applications does not need to be retained, but can rather be analyzed and then archived or deleted, in accordance with legislative requirements for health information.
I suggest that data from consumer health applications effectively utilizes big data analytics capabilities and is a good place to begin investing in big data solutions. An abundance of health data is already available to health organizations; it simply needs to be properly integrated and analyzed. Consumer health data provides high-volume, detailed and frequently updated health information on a variety of populations, ranging from acute care patients to healthy individuals. Big data solutions make it possible to analyze this rich source of data in all its real-life complexity, shifting healthcare organizations from making decisions based on general statistics or small, controlled studies to large-scale and detailed analyses of health outcomes in typical settings.
Big Data, Bigger Outcomes. Lorraine Fernandes, Michele O’Connor & Victoria Weaver. Journal of AHIMA, vol. 83(10), 2012.