Data Mining for Food-Related Illnesses
In a Washington Post article in 2014, Abby Phillip reported that New York’s Department of Health and Mental Hygiene used data mining software with the help of Columbia University and Yelp to scan 294,000 Yelp restaurant reviews for people who had food-related illnesses. According to the article, ‘…red flag words and phrases like “sick,” “vomit,” “diarrhea” and “food poisoning,” were used to flag 900 cases “for further review by epidemiologists.” Many cases of food poisoning go unreported, and this data mining technique was used to locate those cases. Of the 900 cases, the epidemiologists only found “three separate outbreaks that (had) never been reported.” These cases involved 16 people who had food-related illnesses from eating seafood at city restaurants. In the scheme of things involving the identification of outbreaks reported via 311 calls, the three outbreaks found using Yelp was about average.
This data mining technique is an example of what Stephen Goldsmith, Mayor Bloomberg’s deputy mayor of operations, thinks would empower city agency officials to think on their feet (see Anthony Townsend, Smart Cities, p 206). Even though the technology for data mining social media sites for health related issues is new and imperfect, it is improving and does help city agencies get to those hard to find cases by using information technology. It makes our city’s health officials smarter in finding restaurants that employ bad food handling practices that may otherwise go undetected.
