Application of Quantitative Microbial Risk Assessment, Predictive Modeling, and Machine Learning in Food Spoilage Microbial Prediction
學生姓名:
黃鉉喅
指導教授:
凌明沛
學期:
113上
摘 要:
Foodspoilageisamajorglobalchallengeforfoodsafety,withmicrobialgrowthbeingthekeyfactorleadingtoqualityissuesinfoodproducts.Toaddressthisissue,quantitativemicrobialriskassessment(QMRA)ormicrobialpredictivemodelscanbeemployedtoforecastthegrowthofspecificspoilagemicroorganismsinfood.ThepurposeofthisstudyistoexploretheapplicationofQMRA,predictivemodeling,andmachinelearning(ML)intheriskassessmentandfoodspoilagemicroorganisms.ThemethodswillestablishaQMRAframeworkforfoodspoilageandapplytheFoodSpoilageandSafetyPredictor(FSSP)tofermentedfishproductstopredictthegrowthoflacticacidbacteriaandtheeffectsonthefermentationprocess.Additionally,Fourier-TransformInfraredSpectroscopy(FTIR)datawillbecombinedwithmachinelearning,utilizingaSupportVectorRegression(SVR)modeltopredictspoilagemicroorganismsinchickenliverproductsanddiscussthesurvivalofSalmonellaatlowtemperatures.TheresultsshowthatfoodspoilageQMRAcaneffectivelyevaluatetheimpactofdifferentstorageconditionsonfoodspoilage,contributingtoenhancedfoodsafetyandprovidingvaluableinsightsforfoodproducers.TheFSSPpredictionresultsindicatethatliquidsmokeandtemperaturesignificantlyaffectthefermentationprocess,andthepredictivemodelidentifiedtheoptimalfermentationconditions,recommendingafermentationtemperaturecontrolbetween20-25°C.ThecombinationofFTIRandmachinelearningaccuratelypredictedthegrowthofspoilagemicroorganismsinchickenliver,includingthesurvivalandgrowthofinoculatedSalmonellaunderlow-temperatureconditions,demonstratingthepotentialofthismethodforrapidassessmentofspoilagemicroorganisms.Toconclude,thethreestudiesillustratethehighefficiencyofQMRAandFSSPinfoodspoilagepredictionandthestrongpredictiveperformanceofcombiningmachinelearningwithFTIR.Thisapproachcanenhancefoodsafetymanagementbyenablingearlypredictionofmicrobialgrowthinfoodandoptimizingstorageconditionstoextendtheshelflifeoffoodproducts.
黃鉉喅.pdf