Since 2011, Sickweather’s work is a case study in using big data in population health surveillance. The Chief Technology Officer (CTO), Michael Belt offered some perspective in this regard.
Big data has become a major player in public health, spurring new branches of practice such as infodemiology and infoveillance. The World Health Organization reports that by sharing and analyzing information, big data can bridge the gap between clinical medicine and population health. Big data also allows for precision in targeting interventions, which is useful given limited public health resources.
However, it is not quite the utopia one might think. In a complex and emerging field, challenges span logistical, ethical and governance issues. For one, there is always the potential threat to privacy. As the World Health Organization says, “privacy protection is a right and the preservation of public trust is a necessity."
So how does an organization like Sickweather manage this challenge? According to Belt, no personally identifiable information or medically sensitive records are collected. Additionally, once the data are collected, their security is ensured by preventing SQL injections, using secure socket layering and optimal stack and hosting providers.
This happens to be good for business. Belt adds, “As CTO, I see the value from the perspective of the technical teams that engage with our API. I know that they come to us because we are secure, reliable, and have data available in real-time as granular as latitude and longitude pairs, which can be reverse geocoded to any geography that they want to measure." Prior to Sickweather, Belt developed sites for National Institutes of Health (NIH) and Johns Hopkins University, and he has since been a featured speaker at several technology events, including Ernst & Young's Life Sciences Summit.
In ensuring the resilience of big data systems against emergency issues such as outages, the choice of server appears to be critical. The organization chose a hosting provider with a good reputation for uptime. This allows Sickweather to handle a peak activity of 30000 requests per second and to register uptime of 99.99999%. The system also automatically reduces the number of servers during off-peak hours to remain cost-effective.
They learned the hard way. “In our early days, our server (which was purchased in Craigslist for $400 and hooked up to a local ISP) would crash whenever we were featured on the Today Show (which happened 3 times) and other national programs. We wouldn't learn about these media hits until after it was too late and my co-founders would be knocking on my door at 7am to let me know the server was down. I haven't had my phone on do-not-disturb mode ever since," Belt recalls.
What is the future of this field? Belt is confident that it will involve addressing a prolific range of illnesses while being context specific. It is no surprise that this is the vision he shares for Sickweather. In his words, “It seems the direction is heading toward Sickweather being an open developer platform that allows input and output across all diseases and vectors of illness, not just the 30-40 illnesses that we track right now. We are also putting great efforts into localizing our data in different countries and languages to offer a truly global disease surveillance platform.”
submitted by: Ebele Mogo, DrPH - Public Health Consultant