Predicting alcohol use disorder remission: a longitudinal multimodal multi-featured machine learning approach

Abstract

Predictive models for recovering from alcohol use disorder (AUD) and identifying related predisposition biomarkers can have a tremendous impact on addiction treatment outcomes and cost reduction. Our sample (N = 1376) included individuals of European (EA) and African (AA) ancestry from the Collaborative Study on the Genetics of Alcoholism (COGA) who were initially assessed as having AUD (DSM-5) and reassessed years later as either having AUD or in remission. To predict this difference in AUD recovery status, we analyzed the initial data using multimodal, multi-features machine learning applications including EEG source-level functional brain connectivity, Polygenic Risk Scores (PRS), medications, and demographic information. Sex and ancestry age-matched stratified analyses were performed with supervised linear Support Vector Machine application and were calculated twice, once when the ancestry was defined by self-report and once defined by genetic data. Multifeatured prediction models achieved higher accuracy scores than models based on a single domain and higher scores in male models when the ancestry was based on genetic data. The AA male group model with PRS, EEG functional connectivity, marital and employment status features achieved the highest accuracy of 86.04%. Several discriminative features were identified, including collections of PRS related to neuroticism, depression, aggression, years of education, and alcohol consumption phenotypes. Other discriminated features included being married, employed, medication, lower default mode network and fusiform connectivity, and higher insula connectivity. Results highlight the importance of increasing genetic homogeneity of analyzed groups, identifying sex, and ancestry-specific features to increase prediction scores revealing biomarkers related to AUD remission.

Using machine learning to predict heavy drinking during outpatient alcohol treatment

“…Sample heterogeneity can reduce overall classifier accuracy, particularly when the confounder alters the association between the prediction target and other features (Li et al, 2011). Other research using machine learning techniques to predict alcohol use outcomes have stratified models by sex to create more accurate classifiers (Kinreich, McCutcheon, et al, 2021;. The approach we used here (i.e., stratified…”

“…Sample heterogeneity can reduce overall classifier accuracy, particularly when the confounder alters the association between the prediction target and other features (Li et al, 2011). Other research using machine learning techniques to predict alcohol use outcomes have stratified models by sex to create more accurate classifiers (Kinreich, McCutcheon, et al, 2021; Kinreich, Meyers, et al, 2021). The approach we used here (i.e., stratified model development) provides one potential solution to this problem, but there are other methods available (e.g., Li et al, 2011) to address this problem and future research will be necessary to identify the most effective strategy.…”

“…Given the flexibility of these algorithms, they are well-suited to handle questions encountered in behavioral research (Bzdok & Meyer-Lindenberg, 2018). Indeed, these techniques have been successfully applied to address a range of problems in addiction and psychiatry research (Mak et al, 2019), such as predicting hospital readmissions (Morel et al, 2020), predicting substance use disorder treatment outcomes from archival clinical data (Acion et al, 2017), and predicting naturally occurring AUD remission (Kinreich, McCutcheon, et al, 2021;. Other studies have used clinical trial data to build machine learning models to predict treatment outcomes in the context of clinical trials.…”

“…There are crucial sex differences in the factors that maintain harmful alcohol use (Peltier et al, 2019; Walitzer & Dearing, 2006). Recent research using machine learning techniques to predict health outcomes, such as development of alcohol use disorder (Kinreich, McCutcheon, et al, 2021; Kinreich, Meyers, et al, 2021) and heart disease (Segar et al, 2021), has demonstrated the benefits of using demographically stratified models.…”

Occupational factors associated with long-term abstinence among persons treated for alcohol dependence: A follow-up study

“…[ 40 ] Regular employment is the strongest predictor of a good prognosis among alcohol-dependents. [ 41 42 ] Blue-collar workers had higher remission rates than white-collar workers. [ 43 ] More skilled employees were abstinent from AD in an industrial setting, and the reason for maintaining abstinence was the threat of losing the job.…”