Food borne illnesses are becoming increasingly widespread (on Digital Journal Karen Graham regularly alerts readers to the latest threats and recalls). Once a foodborne illness has been reported the rapid detection of the contaminated food is required. Given the complexity of most food distribution systems, this is not an easy task.
For tracking contaminated food, the information collated from grocery scanners can prove invaluable. However, until recently the use of this data has not been attempted. This has changed following a report from IBM Research-Almaden.
IBM scientists have shown how analyzing retail-scanner data from grocery stores in relation to maps and patterns of confirmed cases of foodborne illness will aid investigations. Through modelling data, the science team have shown how, with as little as ten medical-examination reports, an investigation can be narrowed down to 12 suspected food products in less than three hours.
For the computer model, meta-data was used to factor in spatio-temporal data like geographic location and time of consumption. Other factors considered included product shelf life and the characteristics of certain foods for likelihood of being prone to a specific pathogen.
The model proved considerably faster in relation to current scoping techniques, which can take several weeks. Traditional methods are largely based on face-to-face questionnaires with those involved or affected by an outbreak. One of the longest to resolve was a major outbreak of Escherichia coli in Europe in 2011. This took 60 days, with the source being traced to imported fenugreek seeds.
Thermo Scientific (@ThermoSciFood) “Could grocery scanner data speed foodborne illness outbreak tracing?”
Speaking with Controlled Environments, Kun Hu, of IBM, said: Our study shows that Big Data and analytics can profoundly reduce investigation time and human error and have a huge impact on public health.”
The method has been applied to one real-life outbreak in Norway where up to 2,600 different food products were implicated as potential sources. The grocery scanner analysis broke this down to 10 products within a few hours and the the specific product (assuages) identified within a day.
The data approach has been published in the Sigspatial Journal. The paper is titled “From Farm to Fork: How Spatial-Temporal Data Can Accelerate Foodborne Illness Investigation in a Global Food Supply Chain.”
