Kendra Wyant

Research Interests: Machine Learning and Digital Phenotyping

Addiction Research Center: arc.psych.wisc.edu

Research Advisor: John Curtin

Personal Website: Kendra Wyant

Curriculum Vitae

I am interested in machine learning and predictive methods in the context of addictive behaviors.
Broadly, my research focuses on developing models to predict when a drug or alcohol lapse may
occur using an individual’s smartphone data. My current project focuses explicitly on linguistic
features extracted from participants’ SMS data in concert with self-reported context variables.
The overall aim of this project is to identify low-burden measures with enough signal to
accurately predict when someone might lapse and use these models to advance treatment and
intervention for drug and alcohol addiction. For example, through digital phenotyping, we can
collect various passive data sources (via a smartphone application) at a low cost and burden to
the individual. Clinicians can then use these data to implement just in time interventions when
individuals are most likely to relapse, thus creating a more efficient use of treatment resources
and reaching the most vulnerable populations.

Current Projects:
Contextualized daily prediction of lapse risk in opioid use disorder by digital phenotyping

Dynamic, real-time prediction of alcohol use lapse using mHealth technologies