ABSTRACT
Objective To establish a discriminant model and to provide a relatively accurate scientific basis for the early diagnosis of tuberculosis (TB) and detection of the close contacts. Methods Through logistic regression analysis, key factors were selected according to Bayes theory and key factors of TB incidence of the close contacts were screened as well as a discriminant model was established. Results The non-TB incidence discriminant function of the close contacts was described as: Y1= -39.831 (constant) + 1.927 X, (sputum-frequency) + 3.528 X2 (education) + 0.309 X3 (contact time) + 5.893 X4 (evade) + 2.140 X5 (ventilation) + 8.706 X6 (cough) + 30.970 X7 (fever). The discriminant function of non-TB incidence of the close contacts was as: Y2 =-57.875 (constant) + 2.343 X1 (sputum-frequency) + 3.965 X2 (education) + 0.361 X3 (contact time) + 6.296 X4 (evade) + 1.348 X5 (ventilation) + 12.984 X6 (cough) + 36.555 X7 (fever). Conduslon The diseriminant model night be used to contribute to the early diagnosis, early intervention and timely treatment on those close contacts of tuberculosis cases.