Food safety has a major correlation with health related to the public. Machine learning can be a great help for large volume and emerging data sets to enhance the safety of the food supply and minimise the impact of food safety incidents. Pathogen genomes which are food borne and unique data streams, transactional, including text, and trade data, have ample emerging applications initiated by a machine learning approach, like prediction of antibiotic resistance, source related pathogens, and detection of food borne outbreak and also assessment of risk. In this chapter, a gentle introduction of machine learning in the pretext of food safety and a detailed overview of various developments and applications has been enumerated.
TopIntroduction
Public health continues to be jeopardised by food safety. Machine learning has the ability to improve food safety and lessen the effect of food safety accidents by exploiting massive, developing data sets. Foodborne pathogen genomes and novel data streams, such as text, transactional, and trade data, have seen new applications enabled by machine learning, including antibiotic resistance prediction, pathogen source attribution, and foodborne outbreak detection and risk assessment. We give a light introduction to machine learning in the context of food safety, as well as an overview of recent advancements and applications, in this article. Many of these applications are still in their early stages, thus general and domain-specific problems and obstacles connected with machine learning are being identified and addressed.
Foodborne infections continue to pose a significant and long-term threat to public health. Foodborne illness affects 1 in every 6 citizens (or 48 million individuals) each year, resulting in 128,000 admits in hospitals and 3,000 casulties (Scallan et al. 2011) worldwide. The survey of fit and healthy persons world campaign released its vision document in 2010, which included food related safety as a priority (Koh 2010). As per surveillance data from the Foodborne Diseases Active Surveillance Network (FoodNet), not any one of the vision's targets for eradicating almost six major pathogens which are foodborne till 2020 had reached as close as 2019. (Table 1).
Population growth, urbanisation, and globalisation have all occurred, driving and feeding into macro societal changes (Doyle et al. 2015, Phillips 2006). Large number of modulations and advancements related to business of food and supply-chains, comparable to other sectors and industries, have created large chunk of data in nearby years. To increase the safety of the food supply, a variety bunch of data have been investigated in novel methods and at various stages related to the farm-to-table concept. For example, meteorological and terrain data have been checked for guessing contamination of pathogen on farms produce (Strawn et al. 2013), and auditing which is paperless and keeping of record, initiated 1.4 million of periodic measurements which is monthly monitoring of cooking temperature level of rotisserie related to commitment food safety of chickens for in the retail setting (Strawn et al. 2013). (Yiannas 2015).
Table 1. Healthy people 2020 objectives and 2019 preliminary data
Pathogen Detection at the National Centre for Biotechnology Information. The widespread application of WGS in microbiology related to public health safety has spawned the data-driven field of epidemiology which is genomic in nature (Deng et al. 2016).
The developments which are fresh and recent in the field of data science approach and food related safety which have sparked debate over the Big Data domain (Marvin et al. 2017), a term not generally linked to the safety of food. ML has been looked to be viable method for data volume analysis in the case of food and safety to tackle the analytical hurdles posed by the deluge of data.