Data analysis of online information on the net turned out to be able to find food poisoning not reported to the administration
April 2014, the world's largest word-of-mouth website "Yelp"Also landed in Japan, but in the United States of the birthplace of Yelp, experiments were conducted to discover hidden" food poisoning "that was not reported to the administration from the information of the restaurant that was posted, Attempts to utilize the word of mouth for administrative activities are made.
Using Online Reviews by Restaurant Patrons to Identify Unreported Cases of Foodborne Illness - New York City, 2012-2013
Regarding food poisoning in the community, local health centers have a consultation window, but in the United States the telephone consultation window "311" to the administration is the window to receive reports on food poisoning, and citizens are asked about 311 if food poisoning is suspected It is common to telephone / contact the number. However, even if suspected food poisoning, it is thought that there are cases in which the prevention of group food poisoning has failed because the victim does not report, finding information on food poisoning which does not become obvious is a large group food poisoning It is very important to prevent and prevent the spread of damage.
Therefore, in order to detect food poisoning which does not become obviousNew York City Health and Safety Administration (DOHMH)Collaborated with huge word-of-mouth site Yelp, conducted an experiment to discover food poisoning that was not reported to the administration by analyzing the restaurant's information. In addition, this experiment has been conducted for the purpose of finding potential group food poisoning by epidemiologically analyzing word-of-mouth information on the Internet and searching for a method to use it to prevent damage expansion, and also an epidemiologist at Columbia University I am participating.
In the experiment, we first analyzed posting of 294,000 restaurants posted between July 1, 2012 and March 31, 2013 with dedicated analysis software. The analysis software includes those that contain keywords related to food poisoning such as "sick" "vomit" "diarrhea" "diarrhea" "food poisoning" After extracting, two or more identical information are reported, and further written after 10 hours, which is the incubation period of general food poisoning, etc., are referred to as "food poisoning possible It is a screening as sexual posts ".
And the screened posts are analyzed in detail by experts who epidemiologically study food poisoning, and posts that were deemed suspected of food poisoning will be interviewed by the DOHMH via Yelp to the contributors and more detailed analysis Was done and inspected for food poisoning.
In the experiment, 893 posts out of 294,000 reviews were detected as analysis posts "Analysis software posted with potential for food poisoning", after that, 499 cases (56% of the total) resulted from food poisoning as a result of human laboratory check I was certified as a suspicious post. Although the remaining 394 (44%) posts included the keywords themselves, they were postings that deny food poisoning, such as "I thought it looked like Yaba, but I did not get stuck" ~ It is that. In addition, although it was narrowed down to 468 cases suspected as recently occurring food poisoning from 499 postings, it turned out that there was only 15 cases in which reports were raised as food poisoning information at No. 311 It is.
Analysis of the posting in even more detail revealed that 339 out of 468 cases are suffering only by one person, and as a result of examination by experts, it can not be said that the situation is so serious that it can be concluded as food poisoning Excluded as things, 27 out of the 129 remaining posts were certified as "posts suspected of food poisoning" requiring interviews with posters.
For this 27 posts,
1:More than two people are having symptoms
2:Do not take a common meal suspected as another cause of food poisoning
3:People with symptoms belong to different households
Four:That it can be confirmed from the onset of the onset that it is not infected due to other meals
The interview survey was conducted on the check item "Finally, 16 posts were certified as a food poisoning case not reported to number 311.
Besides this experimentUsing TwitterThere are also attempts to detect restaurants that caused food poisoning, etc. By analyzing word-of-mouth information on the Internet, it seems to be said that there is a possibility that you can discover food poisoning that reports are not reported to the administration. In the experiments, it was noted that some of the contributors had the question "Is it food poisoning?", But people who did not know the existence of No. 311 and did not know where to report should be found I deserve it. From here on, the Internet and SNS are rapidly developing, so it seems that the situation that the Internet and SNS is playing the role which was hoped for administrative administration until now can be seen.