Study on the classification of dominant pathogens related to febrile respiratory syndrome, based on the method of Bayes discriminant analysis / 中华流行病学杂志

ABSTRACT

Objective To understand the dominant pathogens of febrile respiratory syndrome (FRS) patients in Gansu province and to establish the Bayes discriminant function in order to identify the patients infected with the dominant pathogens.Methods FRS patients were collected in various sentinel hospitals of Gansu province from 2009 to 2015 and the dominant pathogens were determined by describing the composition of pathogenic profile.Significant clinical variables were selected by stepwise discriminant analysis to establish the Bayes discriminant function.Results In the detection of pathogens for FRS,both influenza virus and rhinovirus showed higher positive rates than those caused by other viruses (13.79％,8.63％),that accounting for 54.38％,13.73％ of total viral positive patients.Most frequently detected bacteria would include Streptococcus pneumoniae,and haemophilus influenza (44.41％,18.07％) that accounting for 66.21％ and 24.55％ among the bacterial positive patients.The original-validated rate of discriminant function,established by 11 clinical variables,was 73.1％,with the cross-validated rate as 70.6％.Conclusion Influenza virus,Rhinovirus,Streptococcus pneumoniae and Haemophilus influenzae were the dominant pathogens of FRS in Gansu province.Results from the Bayes discriminant analysis showed both higher accuracy in the classification of dominant pathogens,and applicative value for FRS.