Drug Abuse Detection via Broad Learning

Abstract

Prescription drug abuse is one of the fastest growing public health problems in the USA. This work develops a broad learning method for Drug Abuse Detection (DAD). In this paper, we propose a new broad learning-based method named ILSTM, short for Improved Long Short-Term Memory, to study the data fusion and prediction from heterogeneous data sources for DAD. The algorithm utilizes the broad learning framework to handle data fusion broadly and information mining deeply simultaneously. Moreover, the effectiveness and prevalence of Holt-Winter inspire our work in the temporal property for DAD.