Establishment of stratified quantitative decision models for measles-related public health emergency events

Objective To establish the indicators for stratifying public health emergency events relating to measles outbreaks, to develop quantitative decision models based on them, and to evaluate the model application. Methods Two models were established based on the C5.0 decision tree algorithm and the polytomous Logistic regression for ordinal response of SPSS Clementine 11.1.1. Results The decision tree model and the criteria for independent variable re-categorization were obtained by the C5.0 algorithm, and the polytomous Logistic regression model for ordinal response was also performed. The evaluation results showed that both models had high accuracy in determining the level of measles-related public health emergency events and predicting outbreaks. Conclusion With the combined application of the decision tree model and the Logistic regression model, determination of measles-related public emergency levels can be automatically achieved. This research was funded by Development and Application of Feasible Information Technology for Disease Prevention and Control (No.2008BAI56B06)