本文旨在建立一种生长/非生长界面模型来预测蜡样芽胞杆菌在环境因子交互作用下的生长概率。选取五株蜡样芽孢杆菌菌株的混合菌株作为研究对象，研究温度、pH、Aw对混合菌株生长概率的交互影响。获得的生长/非生长实验数据用logistic回归方程拟合，建立了环境因子交互作用下蜡样芽孢杆菌生长/非生长界面模型。实验采取部分析因设计方案，选定80%的实验数据用做模型的拟合，20%的数据用做模型的验证。并从已发表的文献中选取30个数据作为测试集，通过比较预测值和观察值来检测已建模型的适用度。实验结果表明，训练集的一致性指数为0.991，验证集的一致率为0.988，说明模型对同类数据预测准确度高；同时模型的R2-Nagelkerke值也较高，为0.949；Hosmer-Lemeshow检验中的χ2=0.012，P=1，logistic回归模型拟合度较高。模型对测试集的预测准确率达83.3%，该模型对所选数据具有较高的预测能力，说明模型具有较广的适用范围。

Bacillus cereus is an opportunistic pathogen, implicated in two main food-borne diseases: emetic syndrome and diarrheal syndrome. The aim of this study was to establish a probabilistic model to predict the growth/no growth conditions or growth probability of B. cereus as a function of temperature, pH, and water activity (Aw). Using a cocktail of five strains of B .cereus, a logistic regression model was chosen to study the effects of different values of temperature, pH, Aw on the growth probability of B. cereus. A fractional factorial design was applied and the experimental data were divided into two parts: 80% data was chosen as model data and 20% data was chosen as validation data. A comparison was made between the predicted and observed values by selecting 30 experimental data from the literature, to test the model. The results showed that the concordance index of the model data was 0.991, while that of the validation data was 0.988. This indicated that the model showed high predictive accuracy for homogeneous data. Additionally, the performance statistics obtained indicated a reasonable goodness-of-?t of the model obtained, mainly owing to the high values of R2-Nagelkerke (0.949) and χ2 = 0.012, p = 1 of the Hosmer–Lemeshow statistic. A high predictive accuracy was obtained (83.3%) with test data, showing a wide range of possible applications.

上海市农业科技成果转化项目（沪农科转字（2015）第2-1号）；农业部引进国际先进农业科学技术项目(2011-Z12)