Computational traceback methodologies are important tools for investigations of widespread foodborne disease outbreaks as they assist investigators to determine the causative outbreak location and food item. In modeling the entire food supply chain from farm to fork, however, these methodologies have paid little attention to consumer behavior and mobility, instead making the simplifying assumption that consumers shop in the area adjacent to their home location. This paper aims to fill this gap by introducing a gravity-based approach to model food-flows from supermarkets to consumers and demonstrating how models of consumer shopping behavior can be used to improve computational methodologies to infer the source of an outbreak of foodborne disease. To demonstrate our approach, we develop and calibrate a gravity model of German retail shopping behavior at the postal-code level. Modeling results show that on average about 70 percent of all groceries are sourced from non-home zip codes. The value of considering shopping behavior in computational approaches for inferring the source of an outbreak is illustrated through an application example to identify a retail brand source of an outbreak. We demonstrate a significant increase in the accuracy of a network-theoretic source estimator for the outbreak source when the gravity model is included in the food supply network compared with the baseline case when contaminated individuals are assumed to shop only in their home location. Our approach illustrates how gravity models can enrich computational inference models for identifying the source (retail brand, food item, location) of an outbreak of foodborne disease. More broadly, results show how gravity models can contribute to computational approaches to model consumer shopping interactions relating to retail food environments, nutrition, and public health.Int. J. Environ. Res. Public Health 2020, 17, 444 2 of 20 success, which is reflected in the large proportion of unsolved cases [5,6]. Low detection rates can at least partly be explained by the enormous complexity of food supply chains and the high proportion of manual work involved in conventional investigation processes. However, emerging technologies and more readily available supply chain data enable authorities to complement investigation processes with computational models [7]. In the case of widespread outbreaks, these data-scientific approaches promise particular utility in outbreak investigations, given the complex task of tracing through the massive food supply network. Data-scientific approaches support the investigation process in multiple parts: (i) detecting that an outbreak is occurring; (ii) identifying the location source of an outbreak at an early stage of the supply chain like a farm or food processor; (iii) identifying the contaminated food item that caused an outbreak; and (iv) investigating other questions relevant to investigators such as identifying the retailer brand source of an outbreak [8].One type of approach (ii) aims to find the location sou...