Francesca Tomasi received her B.A. from the University of Chicago and is now a microbiologist.
6 billion and 4.5 billion: which figure represents the number of people in the world who own a toilet, and which one represents how many own a cell phone? I’ll give you a hint: more people in the world own a cell phone than a toilet.
Today, mobile phone usage encompasses just about every demographic group, and interestingly enough is highest amongst populations most in need of health interventions. As a result, it has a consummate potential to improve worldwide health, especially when coupled with the ever-increasing population also using the Internet. The ability to access and follow previously unreachable communities through mobile and Internet technology swings open many doors in a wide variety of medical fields to improve global healthcare.
Recent years have in fact seen new ideas for exploiting this access to information in meaningful ways. In healthcare, Internet and mobile technologies are starting to be used to organize both individual and aggregate data to facilitate medical intervention and yield predictive health information. Data does not only flow from consumers to large databanks of information, but can now be made available to individuals in forms that are tailored for small towns or valid across the world. Together, these information methodologies are yielding fresh insights into the factors that play a role in diseases and health in general. Organized, tailored information available on-the-go in the palm of one’s hand lends itself to rapid, active changes in behavior or environmental exposures to reduce health risks and optimize both preparedness and overall outcomes in a way we have not yet seen in traditional public health policy. A widespread use of digital monitoring systems adopted alongside traditional methods promises more comprehensible and prompt, community- and individual-oriented healthcare, and the ubiquity of people on the web as well as on mobile phones today is a strong testament to the importance of integrating such technologies in the not-so-distant future.
Of particular concern to public health are emerging infectious diseases, the leading causes of death worldwide and irritants to any community. Driven by socio-cultural, environmental, and ecological factors, epidemics of known, resistant, and previously unrecognized pathogens pose a serious threat to worldwide health. As a result, several monitoring methods have already been established in efforts to prevent, detect, and track rampant infectious agents.
Traditional monitoring of infectious disease is a predominantly passive process sustained and managed to differing degrees around the world. It works through submissions from physicians, laboratories, and other healthcare providers, all of which trickle into a centralized database. The potential inconsistency that arises amongst the varying resources all but guarantees information loss in translation and carries a significant opportunity cost in terms of time and effort. On average, the dissemination of public health data through traditional methods occurs two weeks after the information is initially collected. In the event of a public health emergency, such as the H1N1 pandemic in 2009, two weeks could lead to loss of life. The emerging field of digital monitoring through mobile and Internet technologies can drastically improve the response to infectious disease in a time- sensitive manner.
More specifically, the detection, tracking, reporting, and response systems can be improved by combining predictive analytics, anonymized hospital records, and interactive surveys to analyze current data simultaneously at home and around the world. Interestingly, all of these new approaches used to tackle an illness exploit exactly what they target: behavior. Just as one’s behavior changes in response to new information, the availability of the Internet and associated technologies has altered human behavior related to information seeking, collection, storage, and communication. In the US alone, health-related queries are estimated at approximately 8 million per day. Numerous studies have exploited this behavior on the premise that people who contract a disease will seek information about their condition from the web. Disease incidence can then be estimated by tracking changes in the frequencies of searches for key terms. For example, algorithms have reported accurate estimates of influenza incidence in a given region 1-2 weeks before traditional reports rolled in. Thus, Google Flu Trends was born. Clinical studies have reported positive correlation between Google Flu Trends predictions and emergency room crowding, showing that digital monitoring may become a near real-time tool to help local clinics and hospitals prepare and obtain the proper supplies and resources to deal with sudden surges in people exhibiting flu-like symptoms. Moreover, Google Flu Trends has shown to provide an indication of flu activity up to 4 weeks prior to the release of CDC reports, and with a 97.7% accuracy rate in many locations.
Internet search data has also accurately estimated the incidence of other infectious diseases, such as dengue, a tropical disease spread by mosquitoes. Search query data has provided information that was used to create statistically significantly accurate models of dengue transmission in multiple countries including Bolivia, Brazil, India, Indonesia, and Singapore. In contrast with traditional disease monitoring methods, it is worth noting that Internet resources are free, publicly available, and occur virtually in real-time.
If Internet monitoring of infectious diseases were to complement traditional systems rather than replace them, many predictable drawbacks of stand-alone Internet monitoring systems could also be avoided. Any changes in the status quo of Internet search behavior, such as media-driven interest, could in fact feed false positives in search trends. In addition, usually when someone has had the flu before, they know what to expect; after experiencing the flu or an influenza-like illness once, someone is less likely to Google their symptoms the next time they or a family member get sick. Over time, this could significantly lower the number of searches in a given area and impede the algorithmic predictions. Furthermore, the detection of significant patterns requires large data samples, so that Internet search trends are not very accurate or effective in small towns or for localized outbreaks. People could be searching for information about the flu for reasons other than their own health, which inflates the amount of “predictive” data. These queries could include research, aimless browsing, and sheer interest – I know I Googled “Flu” several times just writing this editorial.
Despite these drawbacks, the fact remains that digital monitoring targets people who are trying to recognize and interpret their illness in addition to those who are seeking treatment. This phenomenon provides public health officials with more data that could be used for preventive and predictive analysis, as traditional methods only target individuals who have already sought treatment and for whom prevention of the disease is no longer really an option. The monitoring of potential pathogens whose hosts experience symptoms but do not seek medical attention also is especially important in some instances. A prevalent symptom of intestinal viruses, for instance, is diarrhea, which many people may experience and choose to leave untreated. Information that does not reach the hospital can provide insight on undercover outbreaks and is otherwise lost without digital surveillance. The importance of prediction lies therefore in its potential to greatly increase prevention, which then not only promotes the control of pathogens, but also allows for a decrease in unnecessary or excessive dispensing of antibiotics. As a result, it may be possible to curb the evolution of resistant, more dangerous pathogens that are then harder to treat than if they had been avoided in the first place.
In order to target a more personalized epidemiological health plan that focuses on an individual’s specific community and individual health rather than state-, nation-, or worldwide disease trends, in the near future we may turn to the Internet by means of more and more mobile technologies in the form of apps and local surveys. These include apps on smart-phones with interactive maps and the ability to report local instances of illness. It may be possible to extend this approach also to SMS-based services for owners of more basic cell phones, which could employ text messaging to report instances of infectious disease and respond with brief simple-text surveys to obtain more information on the potential responsible pathogen and related symptoms in populations where smart phones are not rife.
Apps in the smart-phone world have already demonstrated the effectiveness of collecting and filtering media and user information outside of formal public health channels for improved situational awareness. For example, HealthMap’s app, Outbreaks Near Me, released in 2009, asks users from the general public to contribute reports from their own knowledge and experiences. The aim of the app is to provide local information via an interactive map to users in order to promote behavioral changes that may improve individual and community health. When enough people in a given geographical region have the service installed and use it correctly, the technology carries an incredible potential for one’s own community health status. Right now, the problem lies in a small user base.
Participatory epidemiology through mobile technology carries both great potential and significant problems. Most important among the latter are the concern for user privacy and the verification and filtering of submitted information. It will take a while to figure out effective ways of targeting misuse and guaranteeing consumer privacy. HealthMap for instance has reported its user submissions as coming from three categories: individuals referencing a news article (15%), eyewitness accounts of local events (41%), and personal accounts of illness of either the submitter or a close associate (13%). Everything else is filtered as spam. In any case, a crowd sourcing approach to infectious disease tracking, like Internet-based monitoring, should complement, not replace, traditional efforts.
Once the collection and analysis methods for participatory information have been refined to avoid fake, malicious reports, mobile epidemiological technologies can penetrate a large majority of communities around the world. As long as national and international market conditions remain suitable for the widespread sale of smart-phones and prevailing mobile data-collection technologies develop multi-platform capabilities, we face a future of great opportunities to gather detailed, structured real-time data for public health reporting, especially for underdeveloped regions that currently lack the resources to quickly diagnose, treat, and control disease. A disease can indeed be controlled before it becomes a pandemic, which is particularly vital in the profoundly interconnected world we live in where the spread of pathogens is accelerated by interpersonal communication and travel.
Moreover, when organizations know the symptoms or the specific disease plaguing a town or region in an underserved country, a more proper allocation of aid resources may also take place to replace the current practice known as Duffel Bag Medicine, which involves the sporadic and inexperienced distribution of the pills an organization obtained through donations. This practice is particularly dangerous when dispensing antibiotics or other medications that promote either the development of resistant pathogens or user addiction. Having prompt and prior knowledge allows relief groups to know what type of resources they should obtain, such as relevant vaccinations, symptom treatment, supplements, or, when appropriate, antibiotics. Even in resource-rich countries, individuals informed of an outbreak in their community may either be advised on proactive measures via the app itself to avoid contact with the pathogen, or they may be advised on proper courses of treatment. Often treatment, especially if the cause is a virus, just requires hydration and rest, so people can avoid unnecessary medications, which can account for up to 73% of antibiotic prescriptions, depending on symptoms.
User-oriented digital healthcare technology still has a long way to go. Digital natives’ demand for apps far outpaces the science needed to understand their benefits, risks, and overall impact on health outcomes. Moreover, issues of privacy, confidentiality, regulatory control, and human subject protection are important barriers to research in this area. In addition, Internet-based monitoring, while showing increased potential, still works best in high-income countries that provide ubiquitous Internet access. Efforts in increasing Internet usage in less wealthy countries are essential to allow Internet search monitoring to be applied globally. There is, however, promise to overcome these barriers to realize the full spectrum of digital, real-time health technologies to create safe, scalable, and effective programs. These programs are appealing from a logistical, economical, and epidemiological standpoint thanks to their intuitive and adaptable nature as well as their rapid turnover rates and inexpensive maintenance costs, unlike those of traditional monitoring systems. Because they rely so heavily on behavior, which is known to be quite stochastic, programs need to be flexible, built with models that incorporate several means of collecting information. When these systems are integrated with other sources such as traditional public health records and hospital data, and, for global health trends, with international policies, the tools to understand and address emerging problems will be literally in the palms of our hands.
Brigham and Women’s Hospital. “High rates of unnecessary prescribing of antibiotics continue.”
"Cell phone ownership hits 91% of adults." Pew Research Center RSS. http://www.pewresearch.org/fact- tank/2013/06/06/cell-phone-ownership-hits-91-of-adults/
Dugas, Andrea, Yu-Hsiang Hsieh, Scott Levin, Jesse Pines, Darren Mareiniss, Amir Mohareb, Charlotte Gaydos, Trish Perl, and Richard Rothman. "Google Flu Trends: Correlation With Emergency Department Influenza Rates and Crowding Metrics." Clinical Infectious Diseases 54: 463-469.
Freifeld, Clark, Rumi Chunara, Sumiko Mekaru, Emily Chan, Taha Kass-Hout, Anahi Iacucci, and John Brownstein. "Participatory Epidemiology: Use of Mobile Phones for Community-Based Health Reporting." PLOS: Medicine.
Ginsberg, Jeremy , Matthew Mohebbi, Rajan Patel, Lynnette Brammer, Mark Smolinski1, and Larry Brilliant. "Detecting influenza epidemics using search engine query data." http://static.googleusercontent.com/media/research.google.com/en/us/archive/papers/detecting- influenza-epidemics.pdf
Gunther, Eysenbach. "Infodemiology and Infoveillance: Framework for an Emerging Set of Public Health Informatics Methods to Analyze Search, Communication and Publication Behavior on the Internet." Journal of Medical Internet Research.
"Internet users (per 100 people)." Data from the World Bank. http://data.worldbank.org/indicator/IT.NET.USER.P2/countries/1W?display=default
Lazer, David, Ryan Kennedy, Gary King, and Alessandro Vespignani. "The Parable of Google Flu: Traps in Big Data Analysis." Science 343: 1203-1205.
Milinovich, Gabriel, Gail Williams, Archie Clements, and Wenbiao Hu. "Internet-based surveillance systems for monitoring emerging infectious diseases." Lancet Infectious Diseases 14: 160-168.
S Fox. Online health search 2006: Pew Internet and American Life Project (2006).
"The Global Health Policy Center." Infectious Diseases. http://www.smartglobalhealth.org/issues/entry/infectious-diseases
Young, Roxanne. "Duffel Bag Medicine." (2006)