Predictive Analytics and Machine Learning for the Risk-Based Management of Agricultural Supply Chains

Safe, healthy and resilient food supply chains are essential to ensuring the livelihood and well-being of humans and societies, as well as local and global economies. However, the ability to provide and sustain access to nutritious and safe food continues to be a major concern and a challenge for every country around the world, including developed countries. In fact, a number of serious and global public health risks arise from food supply chains. Two central such risks are adulteration, in which unsafe food is sold for human consumption, and zoonotic diseases (i.e., viruses and diseases that can transfer from animals to humans through food supply chains) such as avian influenza, SARS, and COVID-19. This thesis focuses on food adulteration and zoonotic disease risks, and highlights a variety of use cases and applications in which operations research, machine learning and predictive supply chain analytics can inform the management of these risks through public policy and techno-operational processes. The second chapter focuses on US food imports, and uses network structures of international supply chains (made public from bills of lading) to identify high-risk consignees (importers) at risk for economically motivated adulteration. The third and fourth chapters focus on China's food supply chain, and leverage publicly-posted food safety test results to evaluate risk-based regulatory resource allocations. Specifically, it is shown how risk-based testing can identify more adulteration problems and trace more problems to their source. It is also found for aquatic products that wholesale and wet markets are potentially undersampled by regulators, despite consolidating the riskiest supply from aquaculture farms. The fifth chapter focuses further on Chinese live animal markets, also implicated in numerous zoonotic disease outbreaks, and demonstrates how food adulteration tests and a novel unsupervised clustering algorithm can be leveraged to identify specific markets at risk of spreading zoonotic disease. The sixth chapter uses machine learning to enable more effective development of single-walled carbon nanotube, DNA wrapped, molecular recognition sensors, capable of offering rapid and quantitative detection of adulterants in food. This technology could be extremely advantageous for managing food adulteration risks at wholesale and wet markets in China, where products are quickly sold overnight.

Name thesis:degree_name

Doctoral

Department dc:contributor.department

Massachusetts Institute of Technology. Operations Research Center

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2021

Author dc:creator

• Renegar, Nicholas

Advisor dc:contributor.advisor

• Levi, Retsef

Statement dc:rights

• In Copyright - Educational Use Permitted
• Copyright MIT

Licence dc:rights.uri

• http://rightsstatements.org/page/InC-EDU/1.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/140138

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/140138