AI and dairy

Why did I write the other day about an artificial intelligence dude who I knew 25 years ago, and whose primary application at the time was ensuring elevators in skyscrapers were efficiently dispersed to floors that needed them – oh, and vision?

Because he made the N.Y Times with an hyperbaric headline about making Toronto a high-tech hotbed (he didn’t write the headline) and because his AI basics are making their way into food safety.

Caroline Diana of Inquisitr writes IBM and Cornell University, which primarily focuses on dairy research, will make use of artificial intelligence (AI) to make dairy safe(r) for consumption.

By sequencing and analyzing the DNA and RNA of food microbiomes, researchers plan to create new tools that can help monitor raw milk to detect anomalies that represent food safety hazards and possible fraud.

While many food producers already have rigorous processes in place to ensure food safety hazards are managed appropriately, this pioneering application of genomics will be designed to enable a deeper understanding and characterization of microorganisms on a much larger scale than has previously been possible.

Only a PR thingy could have written this paragraph: “This work could eventually be extended to the larger context of the food supply chain — from farm to fork — and, using artificial intelligence and machine learning, may lead to new insights into how microorganisms interact within a particular environment. A carefully designed informatics infrastructure developed in the IBM Accelerated Discovery Lab, a data and analytics hub for IBM researchers and their clients and partners, will help the team parse and aggregate terabytes of genomic data.”

Better than a poorly designed informatics infrastructure.

Food safety competes with this? FoodPorn, circa 1600s and now, more about status than appetite

Tove Danovich of NPR writes that in the 1600s, when famous still life artist Jan Davidsz de Heem was eating, people showed off their meals with paintings.

food.porn.1600s.jul.16A new study by Cornell’s Food and Brand Lab found that capturing and showing off decadent and expensive meals is a decidedly old-fashioned practice. Brian Wansink, author of Slim by Design, and Andrew Weislogel, a curator at Cornell University’s Johnson Museum of Art, studied 140 paintings of “family meals” from 1500 to 2000 and found that the majority of foods depicted were not part of the average fare. Some of the most likely foods to appear were shellfish, ham and artichoke. For the common classes during the time these paintings were made, Wansink says, more likely items to eat would have been chicken, bread and the odd foraged fruit.

People don’t usually Instagram frozen foods they put in the microwave. Instead, the most successful #foodporn is often an item the photographer laboriously made in the kitchen or found in either an expensive or out-of-the-way restaurant. A recent top #foodporn on Instagram is a photo of seven elaborately decorated eclairs. In the caption the food blogger behind @dialaskitchen compares the Toronto-made pastries to some found a couple years ago, “while at L’atellier de l’éclair in Paris.” Wansink says that today’s social media food posts often attempts to convey that their creator is worldly, adventurous and has money to spare. “None of these things are about food,” he says.

In the paintings, some of the most popular foods are ones that had to be imported or were highly valuable. “It wasn’t Italian paintings that had olives,” Wansink says. “It was the countries that had to import them.” Olives, he points out, are somewhat useless nutritionally and aesthetically. “They look like black marbles,” he says. Even if they are delicious.

Lots of recalls but ‘100 times more likely to detect an outbreak than 20 years ago’

Martin Wiedmann, a professor of food safety at Cornell University’s food science department, told CBC News we really are seeing far more food recalls and outbreaks these days, “But that doesn’t mean our food is less safe. It’s the opposite. What happened over the last 20 years and really accelerated over the last two years is the use of completely new DNA fingerprinting tools to detect disease outbreaks. Today, we are 100 times more likely to detect an outbreak than we were 20 years ago.”

riskHealth officials have developed a system to track the genetic makeup of salmonella, Listeria and E. coli. Once a food-related illness outbreak is identified, scientists can match the DNA from contaminated food with the bacteria making people sick, and potentially trace it to the originating food processing plant.

In light of that long list of recalls, and the fact that we’re detecting more outbreaks, shouldn’t they also be steering us away from salad and cantaloupes? After all, based on the recalls, they might sound like risky foods.

Wiedmann says that’s not really so. He points to the reason we see few cases of issues arising from raw milk consumption as an example of why.

“Much, much fewer people consume raw milk,” he said.

“So we don’t hear much about raw milk outbreaks. But we hear about outbreaks with lettuce, so [people think] lettuce must be less safe. Quite the opposite, because you need to consider the total amount of the food produced — what is your chance of getting sick from eating one of these servings.”

Wiedmann also points out that a recall isn’t the same as an outbreak. In most cases, food recalls are precautionary, and the products haven’t actually made anyone sick.

We call them outbreaks now because we can easily link a specific product in California, for example, with a handful of sick people in separate provinces or states, thanks to the DNA fingerprinting Wiedmann mentioned.

The bottom line, he says, is that those high-risk products health officials advise against, like unpasteurized cheese, are actually riskier than the products making news headlines.

A cheese that’s acceptable in the French countryside isn’t in urban Canada, largely because of our cultural biases.

“The challenge is that risk isn’t binary,” Wiedmann said.

“It’s just not like ‘risk’ or ‘no-risk.’ There’s a gradation… And then somewhere in the middle, someone puts a line. And that line is arbitrary, because no food is risk free.”

How improved veterinary science led to discovery of Salmonella

April 27 was Cornell University’s 150th anniversary. Its charter was signed in Albany in 1865. One of the school’s founders, Ezra Cornell, was a farmer and made veterinary science a priority. This is the story of the career of the first doctor of veterinary medicine to graduate from Cornell.

Salmon_DanielDaniel Salmon was 18 years old in 1868 when he traveled to Ithaca to go to college. Today the College of Veterinary Medicine at Cornell is a sprawling maze of labs, barns and hospitals. Back then it was much simpler.

“There was just one academic building in the very beginning,” says Donald Smith, former dean of the College of Veterinary Medicine. “Cascadilla Hall was there for students and faculty as a residence and there was very little else except a farm, a working farm.

Salmon’s doctorate from Cornell University was the first awarded in the U.S. The bacteria salmonella was named after him, though his assistant Theobald Smith actually discovered it. And Salmon helped found veterinary colleges in Washington, D.C., and Uruguay. He also developed the first federal meat inspection system in the U.S.

When Salmon’s career reached its peak 30 years after arriving at Cornell, it came right as the profession began to change, too. Smith says at first veterinarians were in the cities and made sure horses could get people where they needed to go.

“And the Civil War was a period when they lost probably a million horses and mules from trauma, but mostly from starvation and disease,” says Smith.

The theory that disease is caused by microscopic germs was developed in the 1870s. Then Louis Pasteur first tested his rabies vaccine in 1885. And as Salmon’s career developed, veterinarians began to move out of the cities to work with farmers, treating diseases.

Today, at a state lab run by Cornell, technicians work on samples from farm animals. They load slides with those samples to see what microbes are in there.

Modeling Listeria in leafy greens using micro and big data

I’m not sure what big data means, but it’s a catchy soundbite that is showing up more routinely.

lettuce.skull.noroProf. Martin Wiedmann, food science and technology, has combined the fields of microbiology and big data to better predict disease outbreaks and preserve food safety.

Wiedmann conducted a study focused on Listeria monocytogenes bacteria and related Listeria species — a leading cause of foodborne illnesses and deaths.

“The main goal of the study was to find better ways to determine whether lettuce or similar produce grown in a field are likely to have bacteria on them that could make you sick if you eat the product,” Wiedmann said.

According to Wiedmann, there are about 1,600 cases of Listeriosis annually in the United States, with more than 20 percent of those infections resulting in death.

“It is definitely not your middle-of-the-road food poisoning disease, which makes [Listeria] so important to understand and study,” Wiedmann said.

Normally, raw and unprocessed foods will be preserved with refrigeration or salting to prevent bacterial growth. Listeria can grow under refrigeration temperatures and high salt environments, thus making these typical methods ineffective at killing off the bacteria.

Products that are taken directly from growing conditions and consumed are high risk produce. They are the foods most likely to transmit foodborne pathogens — unless necessary precautions are taken.

“Therefore, it is necessary to make sure the product is safe and free of these pathogens while it is still in the field,” Wiedmann said.

Wiedmann and colleagues collected many samples from various farms in upstate New York.

“In addition to isolating Listeria species from the sample, we also collected Geographic Information System data (GIS), which records the exact location of where the sample was collected,” Wiedmann said. “With this, we can ask questions such as how close was Sample A to water or to a major road?”

Wiedmann uses the data to understand the factors that are conducive to Listeria growth.

Analysis of the data revealed that proximity to water is a major factor of a high risk Listeria presence.

“The analysis of the data allows us to predict high and low risk areas for Listeria and also see whether different types of the bacteria behave differently,” Wiedmann said.

He has also started working on collecting data for Salmonella and Escherichia coli, two other very important foodborne pathogens. According to Wiedmann, the next steps will be to analyze different states and regions in conjunction with different pathogen species to better understand food production and the pathogens that afflict it.

Listeria: On the environmental trail of food pathogens

Tracking one of the deadliest food contamination organisms through produce farms and natural environments alike, Cornell microbiologists are showing how to use big datasets to predict where the next outbreak could start.

listeria4Specifically, the lethal listeria bacterium, Listeria monocytogenes, might be lurking in moist soil, close to open water and near livestock pastures, according to a Journal of Food Protection article, “Geographical and Meteorological Factors Associated with Isolation of Listeria Species in New York State Produce Production and Natural Environments.”

“Due to the complexity of landscapes, it’s basically impossible to know which environments favor the presence of listeria species or other disease-causing bacteria that may contaminate foods,” said Martin Wiedmann, professor of food science and one of five researchers behind the article in the Journal of Food Protection. “We thus took a ‘big data’ approach that combined data from hundreds of bacterial samples from farms and forests across upstate New York with mapping data to identify locations that may favor the presence of specific bacteria.”

The researchers were not surprised to find L. monocytogenes – the deadliest of some 15 listeria species and the cause of listeriosis, which sickened 147 people (and killed 33) with contaminated cantaloupes in 2011 – in farm fields and forests near pastures. Thriving in fecal matter and soil, the rod-shaped bacterium can travel through surface water and other mechanisms to places where human and animal food is grown.

The point of their research was to prove that so-called index organisms (such as other species of listeria collected in samples) can stand in for one bad actor (such as L. monocytogenes) and facilitate detection of microbes of interest. They investigated this by testing whether the same spatial factors that predict the presence of the index organisms also predict the presence of the bad actor.

About 33 percent of samples from New York’s natural environments had some kind of listeria (not necessarily L. monocytogenes), as did 34 percent of samples from produce farms. Samples were taken by scooping sub-surface soil, dragging sterile swabs across land surfaces, and collecting feces and jars of water. Only three of 14 possible geographical factors – soil moisture, proximity to pastures and proximity to water – were highly associated with the isolation of pathogenic (L. monocytogenes) and nonpathogenic (L. innocua, L. seeligeri and L. welshimeri) listeria in produce production environments, Wiedmann and his colleagues reported.

Importantly this study, along with other recent work by this group, provides a blueprint that enables scientists to combine large public datasets that are freely available (like digitized maps and weather data) with lab and testing data to improve food safety and quality.

Other authors of the report are Travis K. Chapin, M.S. ’13; Kendra K. Nightingale, Ph.D. ’05, an associate professor at Texas Tech; Randy W. Worobo, professor of food science in Cornell’s College of Agriculture and Life Sciences; and Laura K. Strawn, Ph.D. ’14, a postdoctoral research associate when the work was conducted and now an assistant professor at Virginia Tech, where she continues to use big data approaches to improve food safety.

The study was funded, in part by National Institute of Food and Agriculture, the U.S. Department of Agriculture, and the USDA’s National Integrated Food Safety Initiative.

Five new Listeria species found; may improve tests

Cornell researchers have discovered five new species of Listeria – including one named for Cornell – that provide new insights that could lead to better ways to detect soil bacteria in food.

To date, of the 10 previously known species of Listeria, only two are pathogenic to humans; Listeria monocytogenes is the main cause of Listeriosis, which causes illness in listeriahundreds – and death in nearly 250 – people each year in the United States through infected deli meats, seafood and produce.

The new study, published online March 5 in the International Journal of Systematic and Evolutionary Microbiology, suggests that all five new species are benign.

The research was part of a larger study led by researchers at Colorado State University and Cornell to examine the distribution of such foodborne pathogens as Listeria, E. coli and Salmonella in agricultural and natural environments. Samples were taken from fields, soil, ponds and streams in New York, Colorado and Florida.

“Doing studies on natural diversity in produce fields helps us develop better and more precise tests to make food safer,” said Martin Wiedmann, Cornell professor of food science and the paper’s senior author.