Examining Use of Pharmacotherapy for Behavioral Support Among Americans with IDD Using Machine Learning

Abstract

ABSTRACT Background Adults with intellectual and developmental disabilities (IDD) often have behavioral support needs that must be met in order to live in community settings. Pharmacotherapies are often used to meet such needs. We explored factors that predict medication for behavioral support among adults with IDD. Methods We integrated three major U.S.- based IDD-specific datasets over multiple years, creating a large dataset. We used machine learning to build and test an algorithm to predict factors associated with the use of pharmacotherapies to address behavioral support needs for adults with IDD. Findings Our algorithm correctly predicted the use of pharmacotherapies with about 79% accuracy. Top predictors of medication use to address behavioral support needs included disruptive behavior, use of medication to treat a mental health condition, and presence of an autism diagnosis. Conclusions Use of behavior plans should be given more attention in conjunction with medication use, especially in relation to managing disruptive behavior.