New recipes

Chicago Using Twitter to Monitor Foodborne Illnesses

Chicago Using Twitter to Monitor Foodborne Illnesses


We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

Chicago is using the power of Twitter to track the spread of foodborne illness across the city.

The Chicago Department of Public Health is taking advantage of Twitter’s prowess to monitor the city for restaurants that may be in need of health inspections, reports PopSci.

The department’s dedicated Twitter bot, and an online complaint form, have thus far helped to identify 133 restaurants for inspections over a period of 10 months. Of those restaurants, 21 failed inspection, and another 33 passed with “critical or serious” violations.

According to a study from the Centers for Disease Control, Foodborne Chicago tracked Twitter messages that included the phrase “food poisoning” to identify “specific instances of persons with complaints of foodborne illness.”

Then, “tweets identified by the algorithm were reviewed by project staff members for indications of foodborne illness (e.g., stomach cramps, diarrhea, or vomiting) from food prepared outside the home. Project staff members provided feedback on whether each tweet fit the criteria, enabling the tweet identification algorithm to learn and become more effective over time.”

Based on the success of the Twitter bot, Chicago is now working with the health departments of Boston and New York to gauge the usefulness of a similar program in those cities.

For the latest food and drink updates, visit our Food News page.

Karen Lo is an associate editor at The Daily Meal. Follow her on Twitter @appleplexy.


Computer model more accurate at identifying potential sources of foodborne illnesses than traditional methods

Boston, MA – A new computer model that uses machine learning and de-identified and aggregated search and location data from logged-in Google users was significantly more accurate in identifying potentially unsafe restaurants when compared with existing methods of consumer complaints and routine inspections, according to new research led by Google and Harvard T.H. Chan School of Public Health. The findings indicate that the model can help identify lapses in food safety in near real time.

“Foodborne illnesses are common, costly, and land thousands of Americans in emergency rooms every year. This new technique, developed by Google, can help restaurants and local health departments find problems more quickly, before they become bigger public health problems,” said corresponding author Ashish Jha, K.T. Li Professor of Global Health at Harvard Chan School and director of the Harvard Global Health Institute.

The study was published online November 6, 2018 in npj Digital Medicine.

Foodborne illnesses are a persistent problem in the U.S. and current methods by restaurants and local health departments for determining an outbreak rely primarily on consumer complaints or routine inspections. These methods can be slow and cumbersome, often resulting in delayed responses and further spread of disease.

To counter these shortcomings, Google researchers developed a machine-learned model and worked with Harvard to test it in Chicago and Las Vegas. The model works by first classifying search queries that can indicate foodborne illness, such as “stomach cramps” or “diarrhea.” The model then uses de-identified and aggregated location history data from the smartphones of people who have opted to save it, to determine which restaurants people searching those terms had recently visited.

Health departments in each city were then given a list of restaurants that were identified by the model as being potential sources of foodborne illness. The city would then dispatch health inspectors to these restaurants, though the health inspectors did not know whether their inspection was prompted by this new model or traditional methods. During the period of the study, health departments continued to follow their usual inspection procedures as well.

In Chicago, where the model was deployed between November 2016 and March 2017, the model prompted 71 inspections. The study found that the rate of unsafe restaurants among those detected by the model was 52.1% compared with 39.4% among inspections triggered by a complaint-based system. The researchers noted that Chicago has one of the most advanced monitoring programs in the nation and already employs social media mining techniques, yet this new model proved more precise in identifying restaurants that had food safety violations.

In Las Vegas, the model was deployed between May and August 2016. Compared with routine inspections performed by the health department, it had a higher precision rate of identifying unsafe restaurants.

When the researchers compared the model with routine inspections by health departments in Las Vegas and Chicago, they found that the overall rate across both cities of unsafe restaurants detected by the model was 52.3%, whereas the overall rate of detection of unsafe restaurants via routine inspections across the two cities was 22.7%.

Interestingly, the study showed that in 38% of all cases identified by this model, the restaurant potentially causing foodborne illness was not the most recent one visited by the person who was searching keywords related to symptoms. The authors said this is important because previous research has shown that people tend to blame the last restaurant they visited and therefore may be likely to file a complaint for the wrong restaurant. Yet clinically, foodborne illnesses can take 48 hours or even longer to become symptomatic after someone has been exposed, the authors said.

The new model outperformed complaint-based inspections and routine inspections in terms of precision, scale, and latency (the time that passed between people becoming sick and the outbreak being identified). The researchers noted that the model would be best leveraged as a supplement to existing methods used by health departments and restaurants, allowing them to better prioritize inspections and perform internal food safety evaluations. More proactive and timely responses to incidents could mean better public health outcomes. Additionally, the model could prove valuable for small and mid-size restaurants that can’t afford safety operations personnel to apply advanced food safety monitoring and data analysis techniques.

“In this study, we have just scratched the surface of what is possible in the realm of machine-learned epidemiology. I like the analogy to the work of Dr. John Snow, the father of modern epidemiology, who in 1854 had to go door to door in Central London, asking people where they took their water from to find the source of a cholera outbreak. Today, we can use online data to make epidemiological observations in near real-time, with the potential for significantly improving public health in a timely and cost-efficient manner,” said Evgeniy Gabrilovich, senior staff research scientist at Google and a co-author of the study.

Funding for this study came in part from the U.S. Centers for Disease Control and Prevention cooperative agreement 1U01EH001301-01.

“Machine-Learned Epidemiology: Real-time Detection of Foodborne Illness at Scale,” Adam Sadilek, Stephanie Caty, Lauren DiPrete, Raed Mansour, Tom Schenk Jr., Mark Bergtholdt, Ashish Jha, Prem Ramaswami, Evgeniy Gabrilovich, online in npj Digital Medicine November 6, 2018, DOI 10.1038/s41746-018-0045-1

Visit the Harvard Chan School website for the latest news, press releases, and multimedia offerings.


Investigating Outbreaks

When a foodborne disease outbreak is detected, public health and regulatory officials work quickly to collect as much information as possible to find out what is causing it, so they can take action to prevent more people from getting sick. During an investigation, health officials collect three types of data: epidemiologic, traceback, and food and environmental testing.

Health officials assess all of these types of data together to try to find the likely source of the outbreak. They take action, such as warning the public, when there is clear and convincing information linking illness to a contaminated food.

Epidemiologic Data:

  • Patterns in the geographic distribution of illnesses, the time periods when people got sick, and past outbreaks involving the same germ.
  • Foods or other exposures occurring more often in sick people than expected
  • Clusters of unrelated sick people who ate at the same restaurant, shopped at the same grocery store, or attended the same event.

Traceback Data:

  • A common point of contamination in the distribution chain, identified by reviewing records collected from restaurants and stores where sick people ate or shopped.
  • Findings of environmental assessments in food production facilities, farms, and restaurants identifying food safety risks.

Food and Environmental Testing Data:

  • The germ that caused illness found in a food item collected from a sick person&rsquos home, a retail location, or in the food production environment
  • The same DNA fingerprint linking germs found in foods or production environments to germs found in sick people

Health officials don&rsquot solve every outbreak. Sometimes outbreaks end before enough information is gathered to identify the likely source. Officials thoroughly investigate each outbreak, and they are constantly developing new ways to investigate and solve outbreaks faster.


Revolutions

Foodborne Chicago finds dodgy restaurants with tweets, and R

If, like me, you've ever had a sandwich from a dubious deli and then been laid up for days afterwards, you know that food poisoning is no trifling matter. In the past, local authorities would only ever learn of such public health issues if they get reported to the authorities by the victim (or the victim's doctor). But that misses the many cases of less serious illnesses that don't involve a doctor or hospital, or illnesses that simply aren't reported to the authorities.

Now, the City of Chicago has found a new way of identifying sources of food poisoning: by analyzing tweets. Foodborne Chicago scans tweets posted in the Chicagoland area, responding to tweets like: "Stomach flu/food poisoning is like eating gas station sushi without the joys of eating gas station sushi" (but ignoring tweets like "It’s really hard to snack while watching Honey Boo Boo. It’s the second best diet to food poisoning."). If you send a such a tweet, you're likely to get a response:

@cheerjoeyniz Sorry to hear you were sick. We can help you by clicking on this link to file a report: http://t.co/jPYs8NxTVw

— Foodborne Chicago (@foodbornechi) April 16, 2013

The system is entirely automated, and uses real-time text analysis implemented with R language to identify those tweets that are about a specific case of food poisoning:

Foodborne searches Twitter for all tweets near Chicago containing the string “food poisoning”. The ingestion service consumes thousands of tweets, storing them in a large MongoDB instance. A collection of classification servers, running R, churn through the collected tweets, applying a series of filters. The tweets are classified using a model that was trained via supervised learning, which determines if the tweets are related to a food poisoning illness or not.

Cory Nissen, the data scientist who implemented the analysis behind the system, shared some of the behind-the-scenes details with me via email. He used an R package called textcat and an algorithm based on n-grams to classify the tweets. The model is trained in such a way as to bias towards sensitivity (at the㻚%+ level) at the expense of specificity (50 - 60%) to better sort true food poisoning reports from "junk" tweets merely about food poisoning. Out of all the tweets in the Chigaco area on any given day, the system flags about 10-20 tweets a day for review, of which just a couple will typically warrant a response to the unwell citizen for followup.

The open-source R code behind the classifier is available on Github. Check out the README file for more technical details behind the implementation. You can also see how the application was presented on Fox 39 Chicago news (starting at the 2:09 mark):


How to Protect Yourself from Foodborne Illness

Foodborne illness is a serious concern, but there’s certainly no need to panic. The good news is you can take some very simple steps and follow a few rules regarding food handling and cooking practices that will greatly reduce the risk of foodborne illness and keep your friends and family safe from Salmonella, and Campylobacter.

When purchasing chicken or pork, for instance, FoodSafety.gov suggests the following tips…

Choose Cold Packages That Are Intact

Make sure the chicken or pork feel cold to the touch with no big tears or holes in the packaging. If possible, put them in a plastic bag so any leaking juices won’t drip on other foods.

Pick Up Meats Last

Make fresh meats the last items to go into your shopping cart. Be sure to separate raw meat from ready-cooked items in your cart. And if you’re getting meat delivered, make sure it goes straight into the fridge upon arrival.

Once you’ve got the meat home there are steps you can take to lessen the likelihood of foodborne illness, too. Fight Bac! Partnership for Food Safety Education is out with helpful food safety, handling, and cooking tips to avoid bacteria. “The basics of clean, separate, cook, and chill will reduce the risk of illness from harmful germs like Campylobacter and salmonella,” says Shelley Feist, executive director of the non-profit organization.

Wash Your Hands—That Means Everyone

Everyone at the gathering and especially those preparing and cooking food should wash their hands with soap and water before and after handling food, not just the grill master.

Keep Food Refrigerated and Cold

Always keep food refrigerated as close to cooking time as possible. For picnics and BBQs keep your (separate) raw meat cooler filled with ice, so picnic perishable foods stay chilled to 40 °F.

Use Separate Plates

Never place cooked food on a plate that previously held raw meat, poultry, or seafood. Be sure to have plenty of clean utensils and platters on hand.

Always Use a Food Thermometer

Measuring the internal temperature of grilled meat and poultry is the #1 way to know your poultry is cooked through and thus safe to eat. Chicken/poultry should always be cooked to 165 degrees Fahrenheit (74 degrees Celsius).

ThermoPro TP03 Digital Instant Read Meat Thermometer, $12.99 on Amazon

A quality, affordable meat thermometer to keep your family safe.

Print a Temperature Chart

Print and hang this helpful guide on your refrigerator or out by the grill, so there is no confusion as to when different meats are cooked through and safe to eat.


Chicago Using Twitter to Monitor Foodborne Illnesses - Recipes

Sharing Stories, Insight, and Experience.

The 10 Food Safety Rules to Follow to Keep Your Family Safe

If you’re anything like my family the kitchen is a sacred gathering place. It’s a place where everyone seems to congregate, engage in conversation, and of course, prepare food. When I think about it, so many of our family and holiday traditions center on food and on food preparation. I can remember at an early age asking my grandmother, “how do you know when the roasted turkey is done?” “How about the meatballs?” Her answers would always be the same, “when the juices run clear [for the turkey],” and, “the meatballs are done around 30 minutes before dinner.” Food safety was never something of concern, nor could grandma answer a question with a firm answer. Of course, I just accepted she was correct – after all, it’s grandma and grandma knows best!

As it turns out, though, food safety is actually very important. Conservative estimates show that nearly 9.4 million illnesses each year are related to foodborne disease. The most common illnesses reported are: norovirus (39%), Salmonella (39%) and E. Coli (3%) with the most common single-food items causing these illnesses being fish, chicken, and then pork. What many people don’t realize is that preventing these foodborne illnesses is actually common sense and very easy-to-do. When it comes to preventing these illness let’s not forget that our children, seniors and immunocompromised family members are the most at-risk for getting sick. We at Fathers of Multiples love sharing great recipes with you. Now we think it’s time we share some food safety tips as well.

With the idea of protecting our loved ones while also providing them safe and delicious foods, here’s a top 10 list of basic food safety principles.

Wash Your Hands

Washing your hands may seem like a simple task however, it cannot be stressed enough. Proper hand-washing is the first defense in proper food safety. Be sure to use hot water and soap. Lather up and scrub both hands: fronts, backs, in between fingers and under the nails. When you’re done use a single-use towel (like a paper towel or a hand-towel dedicated to drying hands). If you’re using a towel that can be machine-washed be sure to replace this towel every other day. Don’t use towels used for wiping the counter or drying dishes to dry your hands.

Use a Food Thermometer

Knowing when your food is properly cooked is critical to food safety. One of the best gadgets you can keep on hand is an instant-read digital thermometer. Don’t go crazy spending tons of money – a simple thermometer can be purchased for under $15.00 and most can be calibrated at home by using simple water and ice. When cooking meats or dishes with mixed meal components the best way to know that it is safe to eat is by the internal temperature. Here is a quick temperature guide when the thermometer is inserted in the thickest part of the meat or dish:

Beef, Lamb, Veal & Fish/Shellfish – 145 0 F

Chicken (ground or bone-in) – 165 0 F

Ground Meat & Pork – 160 0 F

Use Appropriate Refrigerator Temperatures

If food in my refrigerator is kept cold it should be safe, right?

Wrong . How cold matters. Did you know that refrigerators have cold spots (just like ovens have hot spots!)? Here are some quick pointers to remember:

  • Use a thermometer to monitor the refrigerator & freezer temperature
    1. Set the refrigerator to 32 0 F – 40 0 F
    2. Set the freezer to <32 0 F
  • Store temperature sensitive items in the coldest spaces of the refrigerator
    1. Store milk, eggs and dairy products in the back spaces where it’s most cold

Sanitize Daily, Disinfect Weekly (or as needed)

The difference between sanitize and disinfect is that to sanitize is about reducing the probability of a contamination by cleaning and wiping all surfaces whereas disinfection is using a chemical agent to ensure that an area is 100% bacteria free. In most homes proper sanitization is sufficient to prevent foodborne illnesses. It’s typically in public spaces like bathrooms, restaurants etc. that daily sanitization AND disinfection is required. To properly sanitize a surface follow these simple steps:

  • Wipe counter with a dry towel to remove any visible dirt or crumbs
  • Apply a simple unscented-cleaning solution with bleach (you can also use 1 tbsp bleach to 1 gallon of warm, NOT HOT water as a sanitizing solution) for at least 1 minute
  • Using a CLEAN cloth, wipe surface and allow to air dry

Please keep in mind the following:

  • When using store-bought cleaners read all label warning and instructions – they are there for a reason
  • Never combine store cleaners as this can result in harmful fumes (i.e. never combine a bleach-based cleaner with an ammonia-based cleaner)
  • Store all cleaning solutions away from food, locked and out of the reach of children

Change Your Sponge

Plan to change your sponge at least once a week, especially if you can “smell” it. You can microwave to help extend the life of the sponge, but in reality sponges provide a perfect environment for bacterial growth. Food safety starts with having clean spaces to work with, and that can’t happen when using a dirty sponge.

Dish Towel Hygiene

Dish towels are something that many people tend to “get their hands on.” Here are some basic rules to follow:

  1. The dish towel is used for drying CLEAN DISHES only and after a round of “drying dishes” should be allowed to air dry
  2. The dish towel is not for drying hands after washing – use a single use paper towel or a dedicated hand drying towel
  3. All towels (dish and hand towels) should be replaced at least once a week

Prevent Cross Contamination

Cross-contamination is when something that has potential to harbor bacteria (such as raw meat, seafood or poultry) comes into contact with something that is “ready-to-eat.” We can prevent cross contamination by the following:

  • Prepare (cut and store) all high-risk foods first, then wash and sanitize utensils and cutting boards used during the process
  • Wash & dry hands after each time you handle high-risk foods
  • Keep high-risk foods secure and away from ready-to-eat foods
  • Use separate cutting boards for raw meat, seafood and poultry, and preferably use a plastic cutting board that’s non-porous and easily washable
  • Wash and dry all fresh produce before handling

Have Separate Cutting Boards

Cutting boards, especially wooden ones, is a perfect home for bacteria to growth. Using plastic cutting boards is optimal in the prevention of foodborne illness. Be sure to have a few cutting boards, and if possible. You can color code them as to only use one cutting board for one type of food. For example:

  • Use a red plastic cutting board for all raw meat, seafood and poultry
  • Use a green plastic cutting board for all fresh vegetables and produce
  • And maybe a grey one for all ready-to-eat foods that don’t require any cooking (or any additional cooking)

Proper Food Storage

Storing food is critical to preventing foodborne illnesses. Whether the food be in your pantry or in the refrigerator, properly storing food scan do a few things: 1) prevent cross contamination, 2) extend the shelf life of a food and 3) prevent insects from contaminating.

  • Store cereals, grains and dried goods in fully sealable plastic bags or reusable, airtight containers
  • Store meats and high-risk ingredients on a plate or tray in the bottom of the refrigerator (to prevent liquids from dripping down) and within the date stamped on the package
  • Rotate stock and use old ingredients first – read the “use by” and “best buy” dates and rotate your stock (FIFO – First In First Out)
  • Store refrigerated, ready-to-eat foods as high up in the refrigerator as you can
  • Store dairy products in the back of the refrigerator (including milk, cheese, yogurt & eggs!) – it’s the coldest place of the refrigerator
  • Leafy green vegetables that can wilt belong in a high-humidity space (drawer)
  • Store non-leafy vegetables and fruits in a cold place on low-humidity and in original packaging to allow natural gasses to escape, preventing premature rotting

Keep Hot Foods Hot and Cold Foods Cold

There’s a great rule of thumb that many food handlers follow: keep hot foods hot and cold foods cold. There is also a caveat for the hot foods: for no more than four hours at a time. Here are some guidelines to follow especially during those family events where we all know we leave food out for noshing:

  • Hot foods should be kept at or above 140 0 F – anything below for an extended period of time can lead to bacteria growth. Try to use an instant read thermometer every 25 –30 minutes to check for temperature
  • Store cold foods on ice, or in an ice-bin. When ice starts to melt drain the water and replace the ice
  • Proper wrap and store foods within four hours of serving – this is critical

It is one thing to provide delicious, home-cooked meals for our families. It’s entirely another story to provide them with food safety in mind. A little food safety awareness can go a long way, especially when meal planning and preparing for large events. Print this article out and keep a copy in your kitchen as this will be a great, general guideline to follow so that everyone in the family can stay safe and enjoy!

Do you have a question for Jeffrey about the kitchen? Send us a message, and let us know!


Can Twitter and Yelp really help spot a salmonella outbreak?

One of the biggest hurdles to halting foodborne illness outbreaks is spotting the source of the problem -- and spotting it quickly. More often than not, by the time authorities recognize an outbreak of salmonella, listeria or any of the other pathogens that sicken an estimated 48 million Americans each year, it already has had time to spread.

But in recent years, academic researchers and public health officials in New York and Chicago, increasingly have experimented with ways to turn social media platforms such as Twitter and business review sites such as Yelp into early warning systems.

"What makes this useful is the fact that we can get information that's not actually going to public health departments. When people get sick, they usually don't report that," said Elaine Nsoesie, an assistant professor of global health at the University of Washington, who along with colleagues at Boston Children's Hospital and Harvard Medical School has been mining Twitter data for evidence of outbreaks.

Nsoesie said traditional surveillance systems allow health officials to investigate cases only after an individual shows up at a hospital or reports an illness to authorities. But she said the research shows that people often write about their bad experiences on Yelp or complain of their symptoms on Twitter, and that those posts tend to mirror outbreak reports issued by the Centers for Disease Control and Prevention.

Researchers have created keywords (i.e., "food poisoning," "puking," and "diarrhea") as well as particular phrases (i.e., "I went to a restaurant") that are likely to flag posts about foodborne illness. From there, they can determine which cases offer potential signs of a real problem.

"If we can see people actually reporting that they're sick, you could get an early warning that there's an outbreak happening," Nsoesie said. "That's the goal."

The method has yet to prove itself on a broad scale, but some health departments around the country have embraced social media as one potential tool in helping investigators track down the source of outbreaks.

In 2013, officials at the Chicago Department of Public Health, realizing that people suffering from food poisoning often posted on social media about it rather than calling 311 to report the problem, built a program to mine Twitter for complaints about food-related illnesses.

Staff members monitor the potential cases and reach out to the ones that appear relevant, including a link to a form where residents can provide more information about their experience and flag restaurants that might be sickening people:

Between March 2013 and January 2014, the effort identified 2,241 "food poisoning" tweets originating from Chicago and its neighboring suburbs, according to findings published last summer by the CDC. Of those, staff members identified 270 describing specific complaints of foodborne illness.

The nearly 200 reports filed with FoodBorne Chicago, the group overseeing the effort, prompted 133 restaurant inspections. More than 90 percent of those inspections found at least one health violation. But nearly a quarter of the inspections turned up a "critical" violation, such as food not stored at appropriate temperatures.

"It's the real-time nature of this that's great," said Raed Mansour, the project lead for FoodBorne Chicago, recalling a moment last year when the Twitter program spotted three tweets within a single hour about the same restaurant. "We were abl e to mobilize our inspectors right then and there. It's hard to say we prevented an outbreak, but we at least prevented further illnesses."

In New York, officials at the city's Department of Health and Mental Hygiene have worked with researchers at Columbia University to scan reviews on Yelp for unreported cases of foodborne illness. Between July 2012 and March 2013, officials used a software program to analyze nearly 300,000 restaurant reviews and identified 893 that warranted further investigation by epidemiologists.


Computer model identifies sources of foodborne illnesses more accurately

A new computer model that uses machine learning and de-identified and aggregated search and location data from logged-in Google users was significantly more accurate in identifying potentially unsafe restaurants when compared with existing methods of consumer complaints and routine inspections, according to new research led by Google and Harvard T.H. Chan School of Public Health. The findings indicate that the model can help identify lapses in food safety in near real time.

"Foodborne illnesses are common, costly, and land thousands of Americans in emergency rooms every year. This new technique, developed by Google, can help restaurants and local health departments find problems more quickly, before they become bigger public health problems," said corresponding author Ashish Jha, K.T. Li Professor of Global Health at Harvard Chan School and director of the Harvard Global Health Institute.

The study will be published online November 6, 2018 in npj Digital Medicine.

Foodborne illnesses are a persistent problem in the U.S. and current methods by restaurants and local health departments for determining an outbreak rely primarily on consumer complaints or routine inspections. These methods can be slow and cumbersome, often resulting in delayed responses and further spread of disease.

To counter these shortcomings, Google researchers developed a machine-learned model and worked with Harvard to test it in Chicago and Las Vegas. The model works by first classifying search queries that can indicate foodborne illness, such as "stomach cramps" or "diarrhea." The model then uses de-identified and aggregated location history data from the smartphones of people who have opted to save it, to determine which restaurants people searching those terms had recently visited.

Health departments in each city were then given a list of restaurants that were identified by the model as being potential sources of foodborne illness. The city would then dispatch health inspectors to these restaurants, though the health inspectors did not know whether their inspection was prompted by this new model or traditional methods. During the period of the study, health departments continued to follow their usual inspection procedures as well.

In Chicago, where the model was deployed between November 2016 and March 2017, the model prompted 71 inspections. The study found that the rate of unsafe restaurants among those detected by the model was 52.1% compared with 39.4% among inspections triggered by a complaint-based system. The researchers noted that Chicago has one of the most advanced monitoring programs in the nation and already employs social media mining techniques, yet this new model proved more precise in identifying restaurants that had food safety violations.

In Las Vegas, the model was deployed between May and August 2016. Compared with routine inspections performed by the health department, it had a higher precision rate of identifying unsafe restaurants.

When the researchers compared the model with routine inspections by health departments in Las Vegas and Chicago, they found that the overall rate across both cities of unsafe restaurants detected by the model was 52.3%, whereas the overall rate of detection of unsafe restaurants via routine inspections across the two cities was 22.7%.

Interestingly, the study showed that in 38% of all cases identified by this model, the restaurant potentially causing foodborne illness was not the most recent one visited by the person who was searching keywords related to symptoms. The authors said this is important because previous research has shown that people tend to blame the last restaurant they visited and therefore may be likely to file a complaint for the wrong restaurant. Yet clinically, foodborne illnesses can take 48 hours or even longer to become symptomatic after someone has been exposed, the authors said.

The new model outperformed complaint-based inspections and routine inspections in terms of precision, scale, and latency (the time that passed between people becoming sick and the outbreak being identified). The researchers noted that the model would be best leveraged as a supplement to existing methods used by health departments and restaurants, allowing them to better prioritize inspections and perform internal food safety evaluations. More proactive and timely responses to incidents could mean better public health outcomes. Additionally, the model could prove valuable for small and mid-size restaurants that can't afford safety operations personnel to apply advanced food safety monitoring and data analysis techniques.

"In this study, we have just scratched the surface of what is possible in the realm of machine-learned epidemiology. I like the analogy to the work of Dr. John Snow, the father of modern epidemiology, who in 1854 had to go door to door in Central London, asking people where they took their water from to find the source of a cholera outbreak. Today, we can use online data to make epidemiological observations in near real-time, with the potential for significantly improving public health in a timely and cost-efficient manner," said Evgeniy Gabrilovich, senior staff research scientist at Google and a co-author of the study.


The Pilot

Read our in-depth article for more on how the food inspections pilot worked.

In a recently completed pilot program, the city used analytics to improve the process by which health inspectors identify "critical violations" in food establishments, usually related to improper food temperature. Here's how it worked: The city processed relevant data to identify predicting variables associated with violations, developed a model, ran a simulation and then used this forecast to allocate inspections in a way that prioritized likely violators. This data-optimized trial method sped up the process of identifying critical violations by seven days — meaning that restaurant patrons are that much less likely to contract a foodborne illness.

This restaurant inspections pilot is part of a broader data analytics program in the city. In 2013, Chicago was one of five winners of Bloomberg Philanthropies' inaugural Mayors Challenge, a competition that encourages cities to generate innovative ideas to solve major problems and improve city life, and which have the potential to spread to other cities. Chicago received $1 million to construct its SmartData predictive analytics operational platform.

Chicago's success in the food inspections pilot holds great promise for cities across the country to change the way they regulate and ensure public health and safety. Here are few of the most valuable takeaways from Chicago's pilot — essential elements for getting the most out of any data analytics initiative:

Harness open data in creative ways: The days are long past when a city could be viewed as a leader in the open-data movement by simply publishing datasets to increase transparency. To serve citizens, cities must leverage the data they publish online to solve public problems. Chicago's open data portal proved pivotal to the food inspections initiative, offering a data source that was accessible by all parties working on the project.

Think horizontally: Traditional approaches to promote public health are typically confined to a single city department. Yet priorities for health and safety can be much more accurately set by mining and analyzing information from across a broad spectrum of sources.

Embrace non-traditional partnerships: In today's resource strapped environment, cities can get more for less by working with partners in academia, the nonprofit world or the private sector. Chicago exemplified this best practice by partnering with a local nonprofit organization, the Civic Consulting Alliance, and with Allstate Insurance, leveraging the talent of the corporation's data science team.

This pilot is an important step forward in learning to use data to transform the way government operates, making it far more responsive and efficient. Chicago has published the code for this initiative on GitHub and plans to do the same for future pilots in other areas of civic concern through the SmartData platform. That means that other cities will be able to take advantage of this work, helping to create an environment for data-powered innovation.


Statisticians using social media to track foodborne illness and improve disaster response

The growing popularity and use of social media around the world is presenting new opportunities for statisticians to glean insightful information from the infinite stream of posts, tweets and other online communications that will help improve public safety.

Two such examples--one that enhances systems to track foodborne illness outbreaks and another designed to improve disaster-response activities--were presented this week at the 2015 Joint Statistical Meetings (JSM 2015) in Seattle.

Tracking Foodborne Illness Outbreaks:

In a presentation titled "Digital Surveillance of Foodborne Illnesses and Outbreaks" presented yesterday, biostatistician Elaine Nsoesie unveiled a method for tracking foodborne illness and disease outbreaks using social media sites such as Twitter and business review sites such as Yelp to supplement traditional surveillance systems. Nsoesie is a research fellow in pediatrics at Boston Children's Hospital.

The study's purpose was to assess whether crowdsourcing via online reviews of restaurants and other foodservice institutions can be used as a surveillance tool to augment the efforts of local public health departments. These traditional surveillance systems capture only a fraction of the estimated 48 million foodborne illness cases in the country each year, primarily because few affected individuals seek medical care or report their condition to the appropriate authorities.

Nsoesie and collaborators tested their nontraditional approach to track these outbreaks. The results showed foods--for example, poultry, leafy lettuce and mollusks--implicated in foodborne illness reports on Yelp were similar to those reported in outbreak reports issued by the U.S. Centers for Disease Control and Prevention.

"Online reviews of foodservice businesses offer a unique resource for disease surveillance. Similar to notification or complaint systems, reports of foodborne illness on review sites could serve as early indicators of foodborne disease outbreaks and spur investigation by local health authorities. Information gleaned from such novel data streams could aid traditional surveillance systems in near real-time monitoring of foodborne related illnesses," said Nsoesie.

The lack of near real-time reports of foodborne outbreaks reinforces the need for alternative data sources to supplement traditional approaches to foodborne disease surveillance, explained Nsoesie. She added Yelp.com data can be combined with additional data from other social media sites and crowdsourced websites to further improve coverage of foodborne disease reports.

Enhancing Disaster Response by Analyzing Social Media:

As part of a team of statisticians from Statistics without Borders (SWB)--an outreach group of the American Statistical Association--Michiko Wolcott and several colleagues evaluated social media traffic posted during and the days following Typhoon Haiyan striking the Philippines in November 2013 to develop a set of social media analytics best practices for emergency response managers.

The project was conducted in coordination with Humanity Road, a volunteer-based charity that delivers disaster preparedness and response information to the global mobile public before, during and after a disaster. The collaboration led to the development of an informational resource for emergency management professionals titled "A Guide to Social Media Emergency Management Analytics."

SWB and Humanity Road are both members of the Digital Humanitarian Network, consisting of volunteer and nonprofit organizations that support leveraging of digital technology in humanitarian response situations.

Wolcott today presented a summary of SWB's recent work with DHN network organizations, as well as the findings and key recommendations in the guidebook during an invited presentation titled "Worldwide Statistics without Borders Projects: SWB Helping Organizations Make Better Decisions."

The project's overall objective was to analyze the tweets to identify best practices for data handling, identify analysis approaches for emergency response and recommend data management approaches. Important considerations and challenges were identified regarding the use and analysis of Twitter-based data sets for disaster response, noted Wolcott.

"Social media can play a critical role in the dissemination of the information, as well as collection of relevant data during natural disasters. The idea of leveraging social media data such as Twitter is intuitively attractive, given their natural ties to mobile devices with obvious disaster response implications," explained Wolcott.

The guidebook notes there are a number of key considerations to ensure the analysis of social media during a natural disaster is designed to meet the objective. The opportunity for data analysis must be properly and promptly identified, and the disaster response resources and analytical resources must work together to determine how to best house, extract and analyze the data.

Among the recommendations for analysis of social media included in the guidebook are the following:

    1. Relevance--Filtering criteria such as country, keywords, hashtags, geolocation, language, type of posts (e.g., organic vs. retweets) and type of poster (e.g., individuals, relief organizations, news organizations, celebrities, etc.) must be carefully considered based on the analysis objective.

2. Geolocatability--In many cases, the basis of all insight from social media posts is the geolocatability of the tweets. Only a portion of the relevant tweets were geolocated, and, of those, 37% come from the Philippines, 25% from the United States, followed by tweets from Great Britain, Canada and Vietnam. While these results show a global interest and awareness in the event, factors such as the proportion of geolocated tweets and the method of geolocation plays an important part in the decisions regarding geolocation. Capturing the specific motivation for the tweets will depend on the analysis objective, with important implications in the design of data collection to which emergency management professionals and analysts must be sensitive.

3. Language--The particularities of Twitter make language identification challenging--the length of messages, heavy use of hashtags and abbreviations and variations in users' communication styles require considerations beyond straightforward use of standard language identification tools. Furthermore, analyzing social media data in countries such as the Philippines presents additional challenges because the country's residents use several languages and issues such as variations in transliteration can add to the challenges. Emergency management professionals and analysts must be prepared to address these issues.

4. Device vs. impact of the disaster on infrastructure--The penetration and usage for a particular device/platform varies by region and country, which must be taken into consideration. Furthermore, emergency management professionals must be sensitive to interruptions in electricity and communications infrastructure because these may affect the data.

The guidebook also offers a list of questions that will help emergency management professionals start a dialogue about social media emergency management analysis. Broad areas covered in the questions are the handling and storage of data creating a baseline and identifying the type of content and trends and planning the reporting time window, location and language.

In recognition of its work on this project, SWB was honored with Humanity Road's 2014 Da Vinci Award, presented to a patron or contributor who supports the organization's programs.