Public health problems don’t exist in silos and neither do their solutions nor their outcomes. We have plenty of evidence to demonstrate the links between unhealthy environments, unhealthy people and unhealthy economies. These links are not one-way streets; they work in feedback loops, making it difficult to know where to intervene and how. Given this complexity, public health has increasingly been integrating other disciplines to both understand and address some of its biggest challenges. Engineering and computer science can provide us with powerful tools to make sense of complexity. Big Data gives us the ability to collect and manage large amounts of information while Data Science –and in turn application of machine learning and artificial intelligence – is the ability to create models to capture underlying patterns within complex systems and to use them to predict trends. It is important that public health practitioners, researchers and policy makers remain aware of the tools at their disposal to assess and to solve problems, especially as the ethics and safety behind their use is a topic of global discussion.
Take the case of an application created by a start-up company in Kuala Lumpur called Artificial Intelligence in Medical Epidemiology (AIME). This disease-outbreak predicting application uses machine-learning models to predict dengue fever up to 3 months prior to an epidemic. Not only is this prediction model valuable for disease prevention and saving DALYs (disability adjusted life years), but it also saves an estimated 400$ million per country due to the heavy costs and resource utilization associated with poorly predicted management techniques and costs to the health system when an outbreak happens.
Another application dedicated to outbreak mapping is Healthmap, an initiative of Boston Children’s Hospital. This uses data sources like news reports, government websites, social media, clippings and blog entries from frontline health care workers in affected areas, and analyses text looking for appropriate terminology on disease and geography while applying a machine learning algorithm to filter out unrelated information (this algorithm uses feedback from analysts to continually improve its accuracy).
In addition to predicting disease outbreaks, advanced technologies can also be used to address problems at the interface of public health and the environment. Currently, IBM researchers are working on a computer system to predict severity of air pollution in different parts of Beijing several days in advance. This system mines data from different models – using adaptive machine learning to identify the best combination of models to use – to compute the risk associated. The goal is also to eventually offer “what-if” scenario analysis and decision support for emission reduction with specific recommendations on mitigating predicted pollution levels through for example, closing certain factories or imposing driving restrictions. Referred to as cognitive computing, this strategy to pull specific information and recommendations from big data sources has been something in which IBM has been a front-runner (see Watson computer).
Applications of advanced technologies in individual health monitoring are also a hot topic of debate. This year’s World Economic Forum Meeting of the New Champions in China hosted a session on “Machine Learning for Health,” shedding some light on potential applications ranging from prediction of neuro-degenerative disease to complicated search algorithms that can answer personalized medical questions. But like most topics in Artificial Intelligence, there are plenty of ethical questions that need answers like data privacy and access laws, what the data will be used for (insurance coverage implications…), how to optimize the role of people within a machine-influenced context, and working towards equity in reaping the benefits of advanced technologies. In fact, the Vitality Institute has recently launched a public consultation on “Responsibility Guidelines for Personalized Health Technology,” with the objective to develop guidelines for responsible stewardship of personalized health technologies (you have until October 15th to contribute if you’d like!). Access to these technologies by marginalized populations is one of the key components of this effort with considerations around cultural sensitivity, affordability and tools to support technology literacy for users. It may be a long while before personalized health technologies are available to marginalized populations, but already, smart-phone based health technologies are making their way to remote parts of the world through crowd funding to improve access to diagnostic tools (check out Peek Retina eye exam adapter).
Advanced technologies are slowly being added to the public health researcher and practitioner’s toolkit, but they need to be well understood and appropriately rolled out if they are to work as effective, safe, ethical, and sustainable means of responding to some of our greatest challenges. These emerging technologies are not silver bullets, but they can become a powerful means of helping us collect more accurate and timely information, which in turn can lead to more effective preventive measures and improved public health practice, thus supporting the growth of responsive and resilient health systems. Becoming more engaged in discussions and forums around advanced technologies and collaborating with those who develop them can help public health practitioners not only to get to know these tools but to also play an active role in shaping their future. Consider this an invitation to data scientists and engineers for the next Health Systems Research symposium. Strut your stuff.