Our research group sits at the interface of substance use disorder treatment, artificial intelligence, statistics and neuroscience. The overall goal is to leverage new technology to match patients with substance use disorders to optimal treatment. We use both bottom-up (phase I/II clinical trials, single site human subjects research) and top-down (phase III and beyond clinical trial data analytics, electronic medical data and claims databases, mobile devices) methods. Current research projects are in the following areas. Faculty collaborators are at Columbia unless otherwise noted.
1. Addiction Data Science
Individual Level Predictive Modeling of Opioid Use Disorder Treatment Outcome (NIH HEAL Initiative)
This project will develop models to predict individual patient’s risk for relapse (or dropout from treatment) when treated with medications for OUD including methadone, buprenorphine, or extended-release depot naltrexone.
We are currently harmonizing clinical trial datasets from 3 multi-site, NIDA sponsored clinical trials in opioid use disorder and developing a portfolio of projects relating to analytics, harmonization, Common Data Models and related areas.
For more information, see CTN Dissemination Library.
Daniel Feaster, Ph.D. - UMiami
Ying Liu, Ph.D.
Raymond Balise, Ph.D. - UMiami
2. Individual Level Biomarkers for Substance Use and Co-Morbid Disorders
Electroretinogram as a Novel Dopamine Biomarker
We are developing new biomarkers that can predict treatment responses in tobacco use disorder. In particular, we are focusing on the neural circuit in the retina as a window to measure various neurotransmitters. There is an ongoing pilot study for using electroretinogram, which records electrical signals from the retina in response to light flashes, as a biomarker for dopamine release.
Diana Martinez, M.D.
Stephen H. Tsang, M.D., Ph.D.
Ragy Girgis, M.D.
Novel Neuroscience-Based Biomarker Development in Addiction
We are part of the Columbia Addiction Biomarkers Workgroup, and are currently developing several new biomarkers including functional near infared spectroscopy (fNIRS), multi-modal eye tracking, quantitative pain measures, and other technologies. The Data Science Research Group primarily assist in developing the statistical and quantitative methods applied to novel biomarker development projects.
Caroline Arout, Ph.D.
John Mann, M.D.
Elizabeth Sublette, M.D.
3. Methodological Development in Predictive Modeling and Causal Inference in Mental Health
Predictive Modeling of Suicide Behavior in Pharmacotherapy
We are developing new techniques and leveraging data science to make individual level predictions of suicide behavior during the course of treatment of bipolar disorder, major depressive disorder, and co-morbid depression and substance use disorders.
Hanga Galfalvy, Ph.D.
Maria A. Oquendo, M.D. - UPenn
Causal Inference of Assault and Violence In Psychiatric Emergency Rooms and Beyond
We are developing new methodological approaches to apply analytic tools to electronic medical records at NewYork-Presbyterian Hospital to assess whether assualt and violence in the psychiatric emergency room can be predicted and modeled. This is in collaboration with the NYP Value Institute
Ryan Lawrence, M.D.
4. Interventions and Treatment Research in Addiction Science
Technology Enhanced Smoking Cessation Treatment
We are developing new combined mobile technology driven psychotherapy and psychopharmacological strategies for e-cigarettes. This line of research is primarily in collaboration with the the Columbia STARS clinic, which enrolls participants for novel therapeutics devlopment for many Substance Use Disorders.
Christina Brezing, M.D.
Frances R Levin, M.D.
John Mariani, M.D.
Participating In Our Research
You may find and contact us to enroll into one of our current studies through Clinicaltrials.gov:
Electroretinogram: a New Human Biomarker for Smoking Cessation Treatment