Machine learning in the social and health sciences

2023

ECSM

Foodborne illness outbreaks originating in food service establishments (FSEs) are a serious public health concern. There are multiple ways that FSEs can contribute to the propagation of foodborne illnesses. However, not all these risks can be detected by health authorities during official inspections for two reasons. First, health authorities have limited resources, meaning not every FSE can be inspected as often and expediently as necessary. Second, because the food safety and hygiene standards of FSEs are dynamic, the inspection results might represent a snapshot that fails to resemble the current FSE situation. Consequently, it would be helpful to have an early warning system that can allow health authorities to predict the risk levels of FSEs and proactively intervene by demanding that high-risk FSEs take appropriate countermeasures. Considering that customers who feel sick after visiting an FSE may not report their experience to health authorities but instead disclose this information by posting online reviews on social media platforms, this study leverages social media data to predict food safety violations by FSEs. The empirical data for this study derive from two sources: the FSE inspection results from the Department of Health and Mental Hygiene in New York City, which labeled each FSE according to whether it violated food safety regulations, and textual reviews posted by FSE customers on Yelp. Machine learning algorithms were employed to extract textual features from the reviews and construct predictive models. Our results demonstrate that customer reviews posted on social media platforms are valuable for predicting FSE food safety violations. Health authorities can leverage social media data to construct predictive models and deploy them to optimize resource allocation and prioritize interventions to prevent foodborne illness outbreaks originating in FSEs, improving public health. FSE managers can also deploy the models to improve their hygiene standards in response to alerts raised. By taking appropriate countermeasures earlier, they can reduce their risk of financial loss due to negative inspection results.

Section: Discussionmentioning

confidence: 99%

Predicting Food Safety Violations via Social Media to Improve Public Health Surveillance

2023

ECSM

“…Replacement of computationally demanding operations with simpler surrogate models is not a novel concept [40,41], and furthermore, recent advances in machine learning technology even allow the initiation of previously unexplored engineering avenues, such as real-time simulations [42][43][44], and uncertainty quantification combined with optimization [45] of complex physics. This nontraditional approach also has great potential in computational science [46], and possibly provides a unique perspective to understand society [47]. However, there are only a few works that consider the effectiveness of numerical methods.…”

Section: Introductionmentioning

confidence: 99%

An ANN-assisted efficient enriched finite element method via the selective enrichment of moment fitting

Engineering with Computers

Kang²,

Jung

et al. 2023

Enrichment techniques that employ nonconforming mesh are effective in modeling structures with discontinuities because numerical issues regarding mesh quality are avoided. However, the accurate integration of the bilinear and linear forms on the discretized domain, which is required in the standard Galerkin-based finite element method, is computationally expensive due to the complexity of the enriched basis function. In this paper, we present a fast and accurate alternative method of numerical integration using nonlinear regression enabled by a multiperceptron feedforward neural network. The relationship between an implicitly represented geometry and the quadrature rule derived from the moment fitting method is predicted by the neural network; the neural network-based regression model circumvents complex computation and significantly reduces the overall online time by avoiding expensive function evaluations. Through the selected numerical examples, we demonstrate the efficiency and accuracy of the current method, as well as the flexibility of the trained network to be used in different contexts.

An ANN-assisted efficient enriched finite element method via the selective enrichment of moment fitting

Kang²,

Jung

et al. 2022

Preprint

Enrichment techniques that employ nonconforming mesh are effective in modeling structures with discontinuities because numerical issues regarding mesh quality are avoided. However, the accurate integration of the bilinear and linear forms on the discretized domain, which is required in the standard Galerkin-based finite element method, is computationally expensive due to the complexity of the enriched basis function. In this paper, we present a fast and accurate alternative method of numerical integration using nonlinear regression enabled by a multi-perceptron feedforward neural network. The relationship between an implicitly represented geometry and the quadrature rule derived from the moment fitting method is predicted by the neural network; the neural network-based regression model circumvents complex computation and significantly reduces the overall online time by avoiding expensive function evaluations. Through the selected numerical examples, we demonstrate the efficiency and accuracy of the current method, as well as the flexibility of the trained network to be used in different contexts.