Use of a machine learning framework to predict substance use disorder treatment success

Abstract

There are several methods for building prediction models. The wealth of currently available modeling techniques usually forces the researcher to judge, a priori, what will likely be the best method. Super learning (SL) is a methodology that facilitates this decision by combining all identified prediction algorithms pertinent for a particular prediction problem. SL generates a final model that is at least as good as any of the other models considered for predicting the outcome. The overarching aim of this work is to introduce SL to analysts and practitioners. This work compares the performance of logistic regression, penalized regression, random forests, deep learning neural networks, and SL to predict successful substance use disorders (SUD) treatment. A nationwide database including 99,013 SUD treatment patients was used. All algorithms were evaluated using the area under the receiver operating characteristic curve (AUC) in a test sample that was not included in the training sample used to fit the prediction models. AUC for the models ranged between 0.793 and 0.820. SL was superior to all but one of the algorithms compared. An explanation of SL steps is provided. SL is the first step in targeted learning, an analytic framework that yields double robust effect estimation and inference with fewer assumptions than the usual parametric methods. Different aspects of SL depending on the context, its function within the targeted learning framework, and the benefits of this methodology in the addiction field are discussed.

Machine‐learning prediction of adolescent alcohol use: a cross‐study, cross‐cultural validation

“…The superiority of machinelearning algorithms was especially noticeable compared to binary logistic regression, pointing toward a potential overlap of different pathological mechanisms and nonadditivity of the predictive effects of different clusters of predictors [34]. Based on prediction measures the elastic-net algorithm showed the best predictive performance.…”

“…parsimonious linear models, which maximize the effect of relevant predictors and minimize the noise related to highly correlated predictors with low predictive power) [23,30]. Considering the complexity of pathological mechanisms and non-linearity of the predictive effects, these techniques might improve the chances of prediction of pathological trajectories, treatment response and matching patients to the most effective treatments [34]. However, previous substance use machine-learning studies fail to consider algorithms other than regularized regression models.…”

“…For instance, random forests (RF), support vector machines (SVMs) and artificial neural networks (ANN) are being used for prediction in other areas of health sciences [31][32][33]. Considering the complexity of pathological mechanisms and non-linearity of the predictive effects, these techniques might improve the chances of prediction of pathological trajectories, treatment response and matching patients to the most effective treatments [34]. Investigating the comparative performance of different types of machine-learning algorithms also has the potential to provide important insights into the nature of the prediction process [35].…”

Evaluation of Drug Addiction Causes in Punjab Using Machine Learning

“…From the table it is clear that most common outcomes of drug addictions are physical and mental health related, Juvenile crime, unemployment, conflicts as discussed by [1][2][3][4][5]. …”

“…LuraAcion.et al [3] uses machine learning algorithm to predict substance use disorder treatment success and his work is better than [4]. RavneetKaur [4] evaluated drug addiction in Punjab using Fuzzy verdict mechanism.…”