Data Science

The availability of large data sources has changed and expanded addiction science in many areas. Our Division has a growing special interest group in developing a research program in addiction data science, focusing on two areas. First, researchers are increasingly applying new non-parametric, Bayesian and other machine learning methods to newer complex data streams such as large clinical trial datasets, electronic medical records, mobile health data, and administrative databases. These methods allow both the development of predictive and prognostic models as well as dissecting and assessing for the direction and size of causal effects of existing interventions. Second, new methodologies are being developed to specifically and more accurately model data streams that are unique to addiction science: urine drug screens, self-report drug use data, longitudinal and contextual craving and withdrawal symptoms and other ecological momentary assessments. Our Division is actively involved with national and international collaborations such as the National Drug Treatment Clinical Trials Network Artificial Intelligence/Clinical Informatics Workgroup (CTN AI/CI) and the Observational Health Data Sciences and Informatics (OHDSI) program. This program is by definition multidisciplinary and we collaborate with several other departments at Columbia including Biomedical Informatics, Biostatistics, Computer Science and Epidemiology. 

Faculty Conducting Research on Data Science in Substance Use Disorders:
Sean X Luo, MD, PhD
Matis Shulman, MD
A Robin Williams, MD
Edward V Nunes, MD

In Biostatistics:
Melanie Wall, PhD
Ying Liu, PhD
Martina Pavlicova, PhD
Kara Rudolph, PhD