Practical foundations of machine learning for addiction research. Part I. Methods and techniques

Area of Science:

• Computational neuroscience
• Data science in behavioral health
• Machine learning applications in addiction research

Background:

• Machine learning (ML) presents a broad set of methods applicable to addiction research, including identifying individuals with substance use disorders (SUD), analyzing neuroimages, and understanding SUD prognostic factors and genetic underpinnings.
• Despite its potential, ML is currently underutilized within the addiction research community.
• This narrative review aims to bridge this gap by introducing ML tools and concepts to addiction researchers.

Purpose of the Study:

• To provide an introductory overview of machine learning methods relevant to addiction research.
• To facilitate the understanding and adoption of ML techniques by researchers in the field of addiction.
• To highlight the continuum between applied statistics and machine learning.

Main Methods:

• This first part of a two-part review focuses on supervised and unsupervised ML methods.
• Key techniques discussed include linear models, naive Bayes, support vector machines, artificial neural networks, and k-means clustering.
• Examples of ML application in current addiction research are provided, alongside open-source programming tools and best practices.

Main Results:

• The review illustrates the application of various ML techniques within addiction research contexts.
• It emphasizes the commonalities and continuum between applied statistics and machine learning.
• Recommendations for good practices in addiction data analysis, such as rationale for tool selection, sample size calculation, and reproducibility, are offered.

Conclusions:

• This review serves as a primer for researchers to incorporate ML into their addiction studies.
• It aims to expand researchers' analytical toolkits and enhance collaboration.
• The work provides resources for deeper understanding and practical implementation of ML in addiction science.