The increasing globalisation and complexity of food supply chains has made it more vital than ever to develop new tools for maintaining and monitoring food safety. A contaminated raw food or food product can quickly be transported long distances and distributed to many locations or used in numerous types of processed food product. This increases the risk of food safety incidents affecting more people and causing greater economic losses.
As an example of a real-world food safety event, in 2019 an outbreak of E. coli O157:H7 in the US was traced to one grower in California. From the first illness in September, it took nearly two months for the US Food and Drug Administration (FDA) to trace the source — collecting and analysing hundreds of distribution records —issue a public warning and request businesses to recall and withdraw products from the market. One company, alone, had to recall over 75,000 pounds of salad products. By December 2019 the FDA reported that 102 people in 23 states had been infected.
This example demonstrates the need for rapid techniques to identify food contamination to protect people’s health and prevent economic loss by businesses. Although not all in widespread use yet, several techniques using big data have potential to improve food safety.
Big data example 1: Using genomics to detect biological contaminants in food
Advances in whole genome sequencing (WGS) and computing make it possible to identify foodborne pathogens accurately and rapidly. It can also differentiate between different strains so that it is possible to tell if infections in different places — as in the example above — are from the same source.
WGS is rapidly taking over from older analytical techniques due to its greater speed and accuracy. There are many types of instrument, using different techniques. Due to the equipment costs it is mainly found in central government laboratories, universities and businesses providing analytical services. The technology is evolving rapidly, however, reducing processing times and lowering costs.
WGS generates vast amounts of data from reading DNA sequences and requires specialist analytical tools to identify which microorganisms the DNA belongs to. It also requires a database of reference genomes to compare the DNA in the samples with. The Global Microbial Identifier (GMI) project, a consortium of experts from 55 countries, is building a global reference source of genomic information on all known microorganisms — bacteria, viruses, parasites and fungi. This is available to food safety professionals and scientists to identify microorganisms in their samples.
IBM has created the Consortium for sequencing the food supply chain to create WGS tools that detect anomalies in foods that indicate food safety hazards. A project with Cornell University is characterising the genetic features of raw milk throughout the milk supply chain and developing tools to detect any anomalies that could be caused by microbiological contamination during production and transport along the supply chain.
Big data example 2: Food safety databases
Large amounts of data are needed for risk assessment and decision making by food safety authorities. The World Health Organization’s Food safety collaborative platform (FOSCOLLAB) integrates multiple data sources essential for preventing and managing disease outbreaks into a single platform, including:
- GEMS Food contamination database: 7 million records on the levels and trends of contaminants in food and their contribution to total human exposure.
- Chronic Individual Food Consumption database: data on food consumption surveys from 26 countries.
- JECFA Database: details on all chemicals evaluated by the Joint Committee on Food Additives and Contaminants (JECFA).
These and other FAO and WHO databases provide the information needed by food safety professionals and authorities to assess the risks of food contamination by ingredients used and contaminants to consumers of food products.
Big data example 3: IoT and smart sensors for monitoring and tracking shipments
The use of RFID technology for tracking shipments along the supply chain has brought many advantages for food businesses. It reduces the labour involved in handling goods on shipping and delivery, during processing and production, and in warehouse management.
RFID tags can hold far more data on the goods than barcodes, including production data, batch, dates, product variables, weights and sizes. Data can be automatically read and written to the tags using wireless devices, giving complete visibility of processes and improving stock management.
With the addition of more sophisticated connected sensors the system could also collect a range of environmental data as goods are moved through the supply chain, such as temperature — even during cooking — and humidity, dust, dirt, microbes or food spoilage chemicals. The EU MUSE-Tech project developed sensors for real time monitoring of temperature, gases and volatile compounds for quality control in food processing.
As this data is gathered on cloud storage systems it could be analysed with big data analytics tools to extract hidden trends and relationships between environmental or production variables and food quality to indicate ways to reduce the risk of food spoilage and contamination.
Big data example 4: Analysing food-buying patterns
In a food-borne disease outbreak rapid tracing and identification of the contaminated products is essential to minimise the health and economic effects and to identify products to take samples from for analysis.
Food sales data, which is automatically collected by food retail companies, can be used to identify food products that are likely to be the source of contamination. A model of the likelihood of particular foods being consumed at different locations can be constructed using a technique called the likelihood-based approach and mapped against the locations of food-borne disease outbreaks. This can help retailers to quickly identify the contaminated products in their retail outlets.
Big data example 5: Analysing online information
People who have visited restaurants are more likely to report an illness on an online reviews site than to the relevant food safety authority. This was noticed by staff of the New York City Department of Health and Mental Hygiene (DOHMH) which then worked with Columbia University to develop software to analyse online reviews posted on Yelp.
The project identified 468 foodborne illness complaints, of which only 15 had been reported to local authority via the dedicated phone line for reporting illnesses. Follow-up interviews with some of the people posting complaints identified outbreaks requiring further investigation and three restaurants that had multiple food safety violations.
This shows that information posted online by the public on widely accessible social media platforms can be harvested to provide a more efficient way to detect food-borne-disease outbreaks and poor hygiene standards in food outlets than waiting for people to report incidents directly to the authorities.
Using big data in the world of food safety
The increasing digitisation in all aspects of business and society is opening up opportunities to monitor, analyse and improve processes, safety and health.
In the food sector, the use of digital and analytical technologies is creating a traceable safer food supply system, from the producer to the consumer. The examples described above show that the intelligent use of data, data sharing and collaboration are essential to develop these rapidly developing fields so that all can benefit to make a smarter, safer food system.