One of the reasons that arouse more reticence among the population when eating out is the possible lack of hygiene at the restaurant visited. Although fear is completely baseless, because as we have seen time and again in bacteriology testing the kitchens of the restaurants they are more hygienic than household, it is understandable that customers feel well, the end of the day no way to monitor that food preparation is being carried out according to the standards of food safety regulations.
In the recent study published in the journal NPJ Digital Medicine and entitled as "Machine Learning epidemiological: real-time detection of foodborne disease scale ', Adam SADILEK, Ashish Jha and presented the company FINDER tool (detector foodborne diseases in real time) able to use information search and location to warn of potentially dangerous restaurants.
The corresponding author, Ashish Jha, professor K. T. Li Global Health at the School of Public Health T. H. Chan and director of the Institute for Global Health at Harvard highlighted some of the possible applications of this new technological advance. These include, the healthcare savings You obtained by avoiding thousands of entries in the emergency rooms of hospitals in the world (and especially in the US. UU., field development of utility), and support and support provided to restaurants and health inspectors to detect and troubleshoot faster and more efficiently.
How does the system work? Really easy. Everything is based on the macrodatos added by users of mobile terminals that allow access to their positioning via GPS and Search History. In this way he software Subsequent searches can interpret the visit of a local that might indicate that the client has contracted a foodborne disease. For example: "How to relieve stomach pain", "Effects of spoiled food", "Drugs against diarrhea".
Anecdotal discern between coincidences and significant trends, the data are found contrasted with those of other diners. When you enter all information collected a pattern begins to emerge, then you can ensure that there is a health problem in the local object of study with a very high confidence level. Detection is made with a accuracy 85%.
To measure this value two US cities were used. UU., Las Vegas and Chicago. FINDER generated a list of restaurants that could potentially be causing disease and discomfort to customers. That list was, thereupon, provided the competent supervisory authorities, which they investigated 61 FINDER cases proposed by the Nevada city, Y 71 in the megapolis of Illinois.
Of these, a 52.3% presented deficiencies in sanitation and hygiene. A value that doubles the average inspections motu Proprio by inspection agencies and totaling more than 10 000 research.
FINDER far exceeded inspections motivated by complaints from customers themselves in all aspects: precision, scale and latency. The reason why this is so, according to researchers who signed the article, It is that "people tend to blame the last restaurant we visited and thus it is likely that complaint forms are delivered to the restaurant that is not", something that is well documented in the existing scientific literature.
The system is not without limitations, one of the most important is that machine learning or machine learning It presents valuable information only after you have spent a lengthy time interval, equivalent to the incubation period of pathogens or appearance of the first symptoms.
In any case there is a Patent improvement over other systems that rely on data mining. If FINDER compared to the model used in Chicago, which draws complaints directly from Twitter, The new computer model is a 68% more effective. Part of the blame lies, However, the social network, which has earned a bad reputation for complaints and disputes that take place in it.
The potential of the tool is not limited to the catering sector, FINDER can be adapted in the future to do disparate pursuits that undoubtedly will have very positive effects on society. I look forward.