Note: the following entry is a stub. Please help us complete it. 

Problems and Purpose

Launched by the Chicago Department of Public Health (CDPH) in 2013, Foodborne Chicago is an initiative that harnesses real-time transmission of information via Twitter to identify food poisoning.[1] The app uses crowdsourced data drawn from social media (Tweets) to detect evidence of unreported food safety problems and foodborne illnesses that were previously discovered using traditional mechanisms of food inspections. Within its first year, Foodborne Chicago helped hundreds of Chicagoans report food poisoning to the CDPH, resulting in over 240 restaurant inspections conducted by the city - inspections that otherwise may not have happened.[2]

Background History and Context

Foodborne illness is a serious, and underreported, public health problem with high health and financial costs. CDC estimates that each year roughly 1 in 6 Americans (or 48 million people) get sick, 128,000 are hospitalized, and 3,000 die of foodborne diseases.[3] The health and financial costs of foodborne illness are high, costing the US $2-$4 billion annually.[4] However, Chicago residents were not necessarily using the city 311 call center or other modalities to report foodborne illness.

Widespread social media use across the US presented an opportunity for CDPH to harness the power of social media to improve foodborne illness surveillance and ensure accurate reporting. As a result, CDPH partnered with the SmartChicago Collaborative to conceptualize this idea and on March 23, 2013, launched the FoodBorne Chicago application with the goal of improving food safety in Chicago.[5]

FoodBorne Chicago tracks tweets using a supervised machine learning algorithm. This algorithm identifies the keywords of “food poison” within the Chicago area and gets smarter at identifying related tweets as CDPH replies to residents that they suspect have a potential case of food poisoning to report. Similar to “syndromic surveillance,” this Foodborne Illness app uses data drawn from social media to detect evidence of unreported food safety problems and foodborne illnesses that were previously discovered using traditional mechanisms of food inspections.[6]

CDPH is currently working to connect this application with their recently developed predictive analytics model which utilizes open data along with inspection data to identify establishments most likely to fail their next inspection. The US Department of Agriculture, New York City, Boston, and Philadelphia are keen on replicating this system and CDPH has made the code available on GITHUB, an open source platform for sharing innovations.[7]

Majority of cases of foodborne outbreaks never end up getting reported to the local health department anywhere in the country. However, Chicago public health officials noticed that although people might not pick up the phone and call the doctor, they are prone to using Twitter about complaining to the world that they got food poisoning from eating out. The CDPH decided to use this opportunity to improve reporting of foodborne illnesses. They collaborated with SmartChicago to develop an app that sifts through thousands of tweets to identify the ones related to food poisoning.

Within the first year of its launch, FoodBorne Chicago reporting led to unannounced health inspections to at 133 restaurants. The health inspectors found at least one critical violation in 20 percent of those restaurants.[8]

Organizing, Supporting, and Funding Entities

FoodBorne Chicago is a collaborative project between the CDPH and Smart Chicago, a civic organization devoted to improving lives in Chicago through technology. It was first developed by Dr. Bechara Choucair, the city’s commissioner of public health.

Participant Recruitment and Selection

Methods and Tools Used

What Went On: Process, Interaction, and Participation

Influence, Outcomes, and Effects

Approximately 110,000 tweets are sent to the FoodBorne Illness app each day from within the Chicago area. In the first 10 months of use, FoodBorne Chicago identified 2,241 potential tweets about food poisoning from Chicago area tweets.[9] After removing retweets, tweets that did were not about a case of food poisoning (e.g., food poisoning facts), tweets that indicated food poisoning was a result of the Twitter user’s own food preparation, and tweets from Twitter users who appeared to be under 18 years old, there were 270 total food poisoning tweets remaining.[10]

Eight of these 270 tweets (3.0%) indicated a visit to a doctor or the emergency room. Foodborne Chicago staff responded to each of the 270 tweets with an individual message, and 193 reports of food poisoning were submitted through the web form. Nineteen (9.8%) of the reports indicated seeking some sort of medical care. Approximately 80% of the inspections resulted in critical violation citations. As of June 20, 2014, the total number of tweets that have been machine classified by the Foodborne Chicago application is at 3,041. Out of that number, FoodBorne Chicago replied through Twitter to 342 tweets, which resulted in 301 submitted reports to initiate a service request.[11]

Through these efforts Foodborne Chicago has captured the attention of the media (Zagat, Chicago Tribune etc.), but its real value is in bolstering existing food safety measures for the residents of Chicago, and empowering them to take action towards making Chicago safer and healthier city. 

Foodborne Chicago has also published its code to make it easy for other cities to replicate their approach. So far, the local government of metro DC has tried to adapt this app to ensure food safety in their city The US Department of Agriculture, New York City, Boston, and Philadelphia are also keen on replicating this system.[12] [13]

Analysis and Lessons Learned

Using Twitter to collect information on food poisoning could lead to people having privacy concerns. Given the government's access to information, it could potentially chart purchases of stomach medication and, say, cross-reference those to a restaurant or store where other customers also bought such medication a short time after a visit. Yet as long as it's just done by aggregation, and not mining personal information, it could provide a benefit to public health. [14]

It must be noted however, that using social media to track foodborne illnesses is no panacea and could be time-consuming and lead to unpredictable results. Furthermore, the approach is targeted towards a specific demography, usually young adults, who are more well-versed with the use of social media. This is also a group who could write off a bad experience as a bad night and not report the health incident at all. Still, Foodborne gives food investigators a head-start on tracking foodborne illnesses. However, to ensure the restaurants are not vulnerable to misreporting, steps must be taken to ensure accuracy in reporting.[15]

See Also 

Smart Chicago

References

[1] https://www.reuters.com/article/us-chicago-twitter-food-poisoning-idUSKBN0GQ25820140826

[2] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[3] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[4] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[5] https://www.reuters.com/article/us-chicago-twitter-food-poisoning-idUSKBN0GQ25820140826

[6] Foodborne Chicago

[7] Foodborne Chicago

[8] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[9] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[10] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[11] https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6332a1.htm

[12] FoodBorne Chicago

[13] https://www.theatlantic.com/technology/archive/2016/01/predictive-policing-food-poisoning/423126/

[14] https://www.theatlantic.com/technology/archive/2016/01/predictive-policing-food-poisoning/423126/

[15] https://www.dnainfo.com/chicago/20140415/downtown/food-poisoning-tracked-through-twitter-with-foodborne-chicago-app

External Links

http://www.smartchicagocollaborative.org/work/health/foodborne-chicago/

Notes

Lead image: @foodbornechi/Twitter https://goo.gl/4hn6MU

Secondary image: FoodBorne Chicago https://goo.gl/PfBUJ1