In many researches, interaction effects are essential to discover associations or make predictions. It is hard to find the effect of complex interaction but only simple interactions (two-way or three-way). However, the complex interaction between more than three predictors may have strong effect on response. Boolean logic expression is one way to express different complex interactions between multiple binary predictors with “not”, “and”, and “or” logic relations.This situation, for instance, arises when a disease diagnostic is desired based on a group of binary symptoms. It is needed to develop a regression methodology to discover and evaluate the strength of such Boolean logic expressions in regression modeling. Moreover, when the number of symptoms is large, it is time consuming to evaluate every possible combination. An efficient searching algorithm is also required. A lot of times, longitudinal data is collected in public health, which may induce correlations among observations. The purpose of this paper is to develop regression modeling and searching algorithm for logic rules in longitudinal study. The performance will be evaluated in simulation studies and real data analysis.