Digital Epidemiology and Precision Public Health


When the word “precision” is discussed in the health and medical fields, the concept that tends to come to mind is that of precision medicine - which is the use of individual level biomedical data such as genomic information to customize clinical recommendations. While this can no doubt be useful in improving healthcare delivery and ultimately health outcomes, there are many factors that impact the individual’s health before they arrive at the hospital. Ideally, from a public health standpoint, we want to intervene before the individual seeks care, to prevent them from showing up at the hospital.

As illustrated in the health impact pyramid below, the greatest population health impact occurs when interventions target broader levels of causation that require little or no individual effort. While there is a role for individual agency, it is generally more impactful to design conducive social, policy and economic contexts for public health than to focus solely on getting people to change their behaviours. For example, it would be easier to get people to walk if the city is planned to include well-maintained, safe and well-lit walking infrastructure like sidewalks, pedestrian bridges, parks, amongst others. It would be easier to get people to eat healthy if they had access to affordable sources of healthy food rather than if they can only afford fast food and live in a food desert. Our health behaviours cannot be isolated from the contexts we live, work and play in.

Health Impact Pyramid
The Health Impact Pyramid

Thankfully, the field of precision public health aims to intervene on population health challenges at a broader level. Precision public health focuses on breaking down public health data on a granular level to uncover areas of high need and therefore target interventions in a more impactful, equitable and cost-effective way.

Averages do not tell us much. This Lancet Report discusses how global health successes, once broken down by demographic factors such as region, country and cause of death, do not accrue to all. For example, in the same country, Liberia, under 5 mortality rates range from 54.2 per 1000 live births to 120 per 1000 live births. Therefore, if we are serious about achieving the Sustainable Development Goals which have a strong focus on health equity, we need our data at a high level of precision.

This same point has long been made by the World Health Organization and urban health researchers and practitioners who call for the use of data to identify pockets of intra-urban inequity often called “hidden cities”. Breaking down the data, also known as disaggregating the data allows us to identify inequities, to target resources and to make equitable policies. This field will be enabled by the emergence of the field of digital epidemiology, also known as infoveillance and infodemiology. We have discussed the field of digital epidemiology in this white paper.

On digital epidemiology as a catalyst of precision public health, the Centers for Disease Control says that digital epidemiology will allow public health practitioners to accelerate early detection and reporting of outbreaks and non-communicable diseases. Digital epidemiology will also provide new sources of information, new and modern means for public health tracking, as well as new ways to communicate information.

An example of digital epidemiology in action would be Sickweather’s work of scanning sites like Twitter for public health reports and events and then mapping them to the geographic location of each report. Sickweather provides the risk of contagious illnesses per ZIP Code Tabulation Area (ZCTA), while imputing this risk for areas with no social media data. This risk estimate is known as a SickScore and is provided on a scale of 0 to 100. The SickScore is imputed using United States census data, geographical data including the location of areas relative to other ZCTAs. Sickweather compares its machine learning models for computing the SickScore, to examine their relative strength and this model is continually iterated.

This spatial mapping of information can then allow for precision public health. For example, with the flu information gained from Sickweather, we can understand flu trends not only as aggregates but broken down to see how they differ geographically. Additionally, we can map this information to other demographic information to identify healthcare, community health, and policy factors that may be related to the differences in health outcomes from one region to another.

What does this mean for public health? In addition to giving us a more detailed understanding of health outcomes, health risks, and health behaviours in our communities, precision public health allows us to understand how these factors may be influenced by other social, economic and demographic factors. This big picture allows us to inform and/or intervene on the broader level social and policy factors, which influence health outcomes. Equity is at the center of the global public health agenda as seen in the United Nation’s Sustainable Development Goals. Precision public health, facilitated by digital epidemiology, allows us to truly start taking equity seriously.

submitted by: Ebele Mogo, DrPH - Public Health Consultant

← Previous entry