Valerie J. Davidson

A Multifactorial Risk Prioritization Framework for Foodborne Pathogens

We develop a prioritization framework for foodborne risks that considers public health impact as well as three other factors (market impact, consumer risk acceptance and perception, and social sensitivity). Canadian case studies are presented for six pathogen-food combinations: Campylobacter spp. in chicken; Salmonella spp. in chicken and spinach; Escherichia coli O157 in spinach and beef; and Listeria monocytogenes in ready-to-eat meats. Public health impact is measured by disability-adjusted life years and the cost of illness. Market impact is quantified by the economic importance of the domestic market. Likert-type scales are used to capture consumer perception and acceptance of risk and social sensitivity to impacts on vulnerable consumer groups and industries. Risk ranking is facilitated through the development of a knowledge database presented in the format of info cards and the use of multicriteria decision analysis (MCDA) to aggregate the four factors. Three scenarios representing different stakeholders illustrate the use of MCDA to arrive at rankings of pathogen-food combinations that reflect different criteria weights. The framework provides a flexible instrument to support policymakers in complex risk prioritization decision making when different stakeholder groups are involved and when multiple pathogen-food combinations are compared.

Fuzzy control system for peanut roasting

Abstract A fuzzy control system was developed for continuous, cross-flow peanut roasting. A roasting kinetic model based on the combined dynamics of heat transfer and browning reactions was developed. A combination feedforward–feedback scheme was implemented in application software developed by this research group for fuzzy rule-based control. Inputs to the control system included numeric values from process sensors as well as linguistic observations by operators. The control system was tested on a pilot-scale roaster and it successfully maintained roasted peanut colour within an acceptable range despite disturbances in roaster bed depth, roasting air temperature and colour setpoint.

Fuzzy logic applications

Fuzzy logic is a modeling method well suited for the control of complex and non-linear systems. This paper illustrates some of the power of fuzzy logic through a simple control example. For the analytical chemist, fuzzy logic incorporates imprecision from measurement noise as well as from linguistic process descriptions to produce operational control systems.

Fuzzy risk assessment tool for microbial hazards in food systems

A Fuzzy Risk Assessment Tool (FRAT) has been developed for early-stage risk assessment of microbial hazards in food systems. The user defines parameters to describe initial hazard level, potential changes during processing and consumer preparation as well as factors related to consumption and health impact. The inputs are defined in linguistic terms or semi-quantitative levels which are converted to fuzzy values. Interval arithmetic is used to compute exposure and risk. Four examples of microbial hazards in food systems are used to demonstrate features of the tool.

). Food-specific attribution of selected gastrointestinal illnesses: Estimates from a Canadian expert elicitation survey

The study used a structured expert elicitation survey to derive estimates of food-specific attribution for nine illnesses caused by enteric pathogens in Canada. It was based on a similar survey conducted in the United States and focused on Campylobacter spp., Escherichia coli O157:H7, Listeria monocytogenes, nontyphoidal Salmonella enterica, Shigella spp., Vibrio spp., Yersinia enterocolitica, Cryptosporidium parvum, and Norwalk-like virus. A snowball approach was used to identify food safety experts within Canada. Survey respondents provided background information as well as self-assessments of their expertise for each pathogen and the 12 food categories. Depending on the pathogen, food source attribution estimates were based on responses from between 10 and 35 experts. For each pathogen, experts divided their estimates of total foodborne illness across 12 food categories and they provided a best estimate for each category as well as 5th and 95th percentile limits for foods considered to be vehicles. Their responses were treated as triangular probability distributions, and linear aggregation was used to combine the opinions of each group of experts for each pathogen-food source group. Across the 108 pathogen-food groups, a majority of experts agreed on 30 sources and 48 nonsources for illness. The number of food groups considered to be pathogen sources by a majority of experts varied by pathogen from a low of one food source for Vibrio spp. (seafood) and C. parvum (produce) to a high of seven food sources for Salmonella spp. Beta distributions were fitted to the aggregated opinions and were reasonable representations for most of the pathogen-food group attributions. These results will be used to quantitatively assess the burden of foodborne illness in Canada as well as to analyze the uncertainty in our estimates.

Fuzzy models to predict consumer ratings for biscuits based on digital image features

Fuzzy models to recognize consumer preferences were developed as part of an automated inspection system for biscuits. Digital images were used to estimate the physical features of chocolate chip cookies including size, shape, baked dough color, and fraction of top surface area that was chocolate chips. Polls were conducted to determine consumer ratings of cookies. Four fuzzy models were developed to predict consumer ratings based on three of the features. There was substantial variation in consumer ratings in terms of individual opinions, as well as poll-to-poll differences. Parameters for the inference system, including fuzzy values for cookie features and consumer ratings, were defined based on the judgment and statistical analysis of data from the calibration polls. The two fuzzy models that gave satisfactory estimates of average consumer ratings are: the Mamdani inference system based on eight fuzzy values for consumer ratings; and the Sugeno inference system developed using the adaptive neurofuzzy inference system algorithm.

Hospitalization and deaths for select enteric illnesses and associated sequelae in Canada, 2001–2004

This paper describes morbidity and mortality parameters for Campylobacter spp., Salmonella spp., enterohaemorrhagic Escherichia coli, Listeria spp., norovirus infections and their primary associated sequelae [Guillain-Barré syndrome (GBS), haemolytic uraemic syndrome, reactive arthropathies and Reiters syndrome]. Data from a period of 4 years were obtained from three national databases to estimate percentage of reported cases hospitalized, mean annual hospitalization incidence rate, frequency of hospitalization by age and sex, and number of deaths. The length of hospital stay, discharge disposition, hospitalization age, and number of diagnoses per case were also extracted and summarized. In addition, we estimated that each year in Canada, there are between 126 and 251 cases of Campylobacter-associated GBS. This study provides morbidity and mortality estimates for the top enteric pathogens in Canada, including their associated sequelae, which can contribute to the quantification of the burden of illness.

A Multi-Factorial Risk Prioritization Framework for Food-Borne Pathogens

To lower the incidence of human food-borne disease, experts and stakeholders have urged the development of a science- and risk-based management system in which food-borne hazards are analyzed and prioritized. A literature review shows that most approaches to risk prioritization developed to date are based on measures of health outcomes and do not systematically account for other factors that may be important to decision making. The Multi-Factorial Risk Prioritization Framework developed here considers four factors that may be important to risk managers: public health, consumer risk perceptions and acceptance, market-level impacts, and social sensitivity. The framework is based on the systematic organization and analysis of data on these multiple factors. The basic building block of the information structure is a three-dimensional cube based on pathogen-food-factor relationships. Each cell of the cube has an information card associated with it and data from the cube can be aggregated along different dimensions. The framework is operationalized in three stages, with each stage adding another dimension to decision-making capacity. The first stage is the information cards themselves that provide systematic information that is not pre-processed or aggregated across factors. The second stage maps the information on the various information cards into cobweb diagrams that create a graphical profile of, for example, a food-pathogen combination with respect to each of the four risk prioritization factors. The third stage is formal multi-criteria decision analysis in which decision makers place explicit values on different criteria in order to develop risk priorities. The process outlined above produces a ‘List A’ of priority food-pathogen combinations according to some aggregate of the four risk prioritization factors. This list is further vetted to produce ‘List B’, which brings in feasibility analysis by ranking those combinations where practical actions that have a significant impact are feasible. Food-pathogen combinations where not enough is known to identify any or few feasible interventions are included in ‘List C’. ‘List C’ highlights areas with significant uncertainty where further research may be needed to enhance the precision of the risk prioritization process. The separation of feasibility and uncertainty issues through the use of ‘Lists A, B, and C’ allows risk managers to focus separately on distinct dimensions of the overall prioritization. The Multi-Factorial Risk Prioritization Framework provides a flexible instrument that compares and contrasts risks along four dimensions. Use of the framework is an iterative process. It can be used to establish priorities across pathogens for a particular food, across foods for a particular pathogen and/or across specific food-pathogen combinations. This report provides a comprehensive conceptual paper that forms the basis for a wider process of consultation and for case studies applying the framework.

Fuzzy predictor for fermentation time in a commercial brewery

Abstract A study was conducted at a commercial brewery to investigate the feasibility of an expert-system control strategy that used fuzzy logic to make inferences based on uncertain process information. The predictor made an initial estimate of fermentation time based on the yeast variables (pitching rate, pitched volume and viability). The predicted fermentation times were within 24 h of the actual fermentation time for 9 batches in the validation data set of 13 process histories. A second rule set followed the pH and specific gravity during fermentation and predicted the level of vicinal diketone (VDK). When the VDK level reached a threshold level, a third rule set updated the estimate of time required to complete the fermentation. Seven data sets were used to validate the predictions based on VDK level and 4 estimates were within 24 h of the recorded fermentation release times. All 7 predictions were within 32 h of the actual release time and 6 of the 7 estimates indicated that the batches were held longer than necessary.

Fuzzy methods for automated inspection of food products

Automated product inspection is of considerable interest to food manufacturers since human inspectors currently perform a substantial amount of on-line inspection. At a low-level of information processing, machine vision offers advantages of objective and consistent assessment. However machine vision systems are frequently used for grading and quality control. In these applications, it is necessary to integrate a number of physical features to make an inference about overall quality that is consistent with consumer judgements. The work presented in this paper focuses on quality assessment of chocolate chip cookies based solely on visual features. Digital images were used to define physical characteristics of cookies produced on a commercial bakery line. Consumers were asked to rate typical cookies on a line scale. A number of fuzzy systems were developed to make quality control decisions based on features extracted from digital images. Results from two fuzzy systems are compared to consumer results from a validation test.