Data Tracking Applications for Coronavirus–Bethel Kifle

Updated: Jan 21, 2021

As the July 4th hype wanes, America begins to brace for yet another unwelcome spike in coronavirus cases. Despite seeing declining cases in most European and Asian countries that once surpassed America in case numbers only a couple months ago, America is yet to catch up with the advancements made by the rest of the world. And still, the worst is yet to come. Dr. Anthony Fauci, the American immunologist serving as head of the National Institute of Allergy and Infectious Diseases, referenced a “disturbing surge” in coronavirus cases even in late June.

The silver lining to this bleak cloud, however, lies in groundbreaking and - not so - new technological advancements in forecasting disease outbreaks developed by research teams under the US Center for Disease Control and Prevention.

One of these teams, proving to be accurate and successful in their endeavors for many years, located in the heart of Silicon Valley, is Carnegie Mellon University’s own machine-learning department. Annually, Roni Rosenfield, head of the department, and his team work to update and adapt their influenza – longhand for the flu– forecasting that has achieved near perfection in determining trends in influenza cases in the past (the team was even asked by the US Center for Disease Control and Prevention to “...lead the design of a community-wide forecasting process”).

It is a known fact that the coronavirus and the flu differ in many categories. Additionally, Rosenfield states that forecasting and pinpointing patterns of the coronavirus is “significantly harder” than those of other diseases such as the flu and worried if his predictions would be “...accurate—and, thus, whether they would even be useful”. Despite this, his team decided to take the risk in hopes of a beneficial outcome.

During flu season, Rosenfield and his team uses two methods to forecast the rises and falls of the flu. The first is called nowcast and is designed to predict the amount of people with the flu by gathering data from disease prevention organizations, flu-related searches on web search engines, social media activity, Wikipedia articles, important retail sales, and much more. This gathering of loads of data, also known as “big data”, allows Rosenfield’s team to access and assess data trends and patterns. In addition, with the use of machine learning to quickly track and reveal these patterns, predictions can be made accurately and in real time.

The second method, known as proper forecasting, is designed to predict the amount of people who will possibly contract the flu in the future. Using a mixture of nowcast and empirical data from years and years of research, proper forecasting utilizes a form of data tracking called crowdsourcing. This form of gathering data allows all new and vital information to be tracked within a secure database through the extraction of services from large amounts of people – usually volunteers. With the added use of more aggregate predictions, Rosenfield’s method of pinpointing the trends of the flu is nearly perfect and gives hope to a new preventive method of tracking the coronavirus.

We know that the flu and the coronavirus are different. But with the algorithms we already have in place from past data accumulations in combination with big tech companies willing to offer their hand in data tracking and reports, Rosenfield’s team at CMU is hopeful. The prevention phase of the United States’ fight with the coronavirus is rapidly approaching – although a bit later than her western democratic counterparts – and Rosenfield is ready to take part in it. “I can do the best I can now,” he says. “It’s better than not having anything.”

19 views0 comments

Recent Posts

See All