November 15, 2016 | Written by: Laxmi Parida
Share this post:
Thanksgiving in the United States, like other holidays around the world, places food at the center of a joyful celebration — and at the forefront of our minds.
As we plan, shop, prepare, cook and share our celebratory meals, we are careful to include not only flavors that our loved ones enjoy, but also foods that are safe.
We avoid allergens, note expiration dates with extra vigilance, and heed manufacturer recalls on contaminated products. This is common sense, but food safety is still a serious health and economic challenge, at Thanksgiving and every day of the year.
In the United States alone, 1 in 6 Americans becomes ill from food or beverages contaminated with foodborne sicknesses each year, and 3,000 Americans will die annually of these, estimates the Centers for Disease Control.
These illnesses cost U.S. individuals and organizations $15.6 billion per year, according to U.S. Department of Agriculture statistics. American food is also being recalled at increasingly higher rates. Between April and June of this year, the U.S. Food and Drug Administration recalled 80 times more food than it did than in the first three months of 2016.
The latter statistic may be more comforting than disturbing. Because testing methods for contamination and supply chain processes continuously evolve to become more efficient, we are able to identify unsafe foods before they reach the table and before our loved ones ingest them.
As a scientist — and a cookbook author — I am on a mission to improve these statistics.
My team at IBM Research is working on a data-driven, predictive approach that will identify systemic problems in food supply chains. Our goal is to use algorithms to discover anomalies within massive amounts of data on plants, livestock, bacteria and other genetic and biological organisms on farms and within other food supply environments.
This algorithmic-based approach to food safety offers new advantages in its admittedly ambitious scale, scope and prescience. Previously, food safety efforts were reactive, based on identifying what has been making patients ill after authorities identify patterns of illnesses or deaths from meat or vegetables tainted by listeria, salmonella, e coli or other pathogens.
Our algorithmic approach will identify danger before it strikes.
Our first steps toward an algorithmic approach to food safety are not easy. My team is creating complex databases of food supply information to find potential “bad guys” in terms of the pathogens or genetic abnormalities that contribute to unsafe food conditions.
We can draw some parallels with the FBI CODIS database of known bad guys. But our database is even more complex, because we are capturing biological information of all food sources on farms and bacterial presence within food supply environments — good as well as bad.
We are even tracking microbes that mutate rapidly into harmful pathogens, sometimes into 100 or more similar versions of the original microbe, but retain most of the genomic signatures of a harmless organism while being actually harmful.
Here’s how algorithms work: they work in lock-step with data and carefully curated databases like ours. Algorithms learn to foresee danger, even when there are no obvious signs of dangerous food, even when a crop or dairy source is technically safe. Furthermore, algorithms do so at a pace that never slows down, at a scale that only can grow.
We used a similar algorithmic approach when we helped to create a way to offer personalized cancer treatments for patients. We used IBM Watson – an organized, intelligent and automated system that ingests large amounts of text (such as millions of medical documents).
This cognitive system works hand-in-hand with our algorithms that capture the current understanding of the pathology of cancer. As a result, doctors can now offer more informed diagnoses and treatments, on a larger scale than before. This method of tackling an urgent health problem via algorithms inspired us to apply algorithms to food supply safety, too.
The true beauty of the algorithmic approach is that it can adjust. Beyond food and science, my other passion is tango dancing, and I see a parallel between tango and algorithms. Tango partners alter their impromptu movements to arrive at compelling, real-time choreography; algorithms constantly adjust and learn.
Our algorithms will help food suppliers predict and re-choreograph their supply chains to avoid producing dangerous food before a single patient suffers, while also avoiding any wastage of precious food.
Laxmi Parida, Ph.D., is a Distinguished Research Staff Member and Manager of the Computational Genomics Group at IBM Research. She will be sharing more about her research at TED@IBM on Nov. 15th. Learn about TED@IBM and how you can get involved here.
This story first appeared on Forbes.