Category: Suicide and Self-Injury

Symposium

Symposium 35 - Exploratory Data Mining in Clinical Research: Demonstrating Applications in Self-Injurious Thoughts and Behaviors

Friday, November 17
12:00 PM - 1:30 PM
Location: Cobalt 502, Level 5, Cobalt Level

Keywords: Self-Injury | Suicide | Statistics
Presentation Type: Symposium

Self-injurious thoughts and behaviors (SITBs), such as suicidal ideation and non-suicidal self-injury (NSSI), are of great public health concern. Not only do these experiences occur in high rates among both adolescents and adults (e.g., 9-25%; Claes et al., 2014; Nock et al., 2008), but also represent high risk for suicidal behaviors (Victor & Klonsky, 2014). Our ability to accurately predict SITBs is limited (Franklin et al., 2016); however, the use of exploratory data mining (EDM) analytic techniques may be one way to improve our predictive ability. The field of EDM (McArdle & Ritschard, 2013; i.e., machine learning) has popularized statistical methods that allow researchers to explore nonlinear relationships, a limitation of traditional statistical methods, and the consideration of interactions between predictors, which is often the norm versus the exception in clinical research. Given the advantages of EDM methodology, the goal of this symposium is to provide an overview of EDM and its application in clinical research. The symposium reaches this goal by bringing together researchers who have an expertise in EDM and those who can demonstrate the utilization of these methods in SITBs research.


 The symposium will begin with a talk providing an overview of EDM methods that are becoming increasing popular in clinical work to predict a single outcome, decision trees (and their extensions) and regularization methods (Jacobucci).


The next three talks will demonstrate applications of these methods as a novel approach to better understand SITB risk. Evan Kleiman (Kleiman, Jacobucci, Ammerman, Turner, Beale, Fedor, Huffman & Nock) will present ecological momentary assessment data showing how the EDM methods of random forests and regression trees were used to identify “self-hatred” and “desperation” (in addition to their associated cut scores) as variables of importance in identifying current suicidal ideation. Taylor Burke (Burke, Jacobucci, Ammerman, Hamilton, & Alloy) will then extend these methods to the prediction of recent and former NSSI, finding anxiety symptomatology as highly important for NSSI prediction, and depressive symptomology and interpersonal vulnerabilities as important for the prediction of recent NSSI. Finally, Brooke Ammerman (Ammerman, Jacobucci, Turner, Serang, & McCloskey) will use a multivariate EDM technique, structural equation modeling trees, to improve the quantification of NSSI severity by identifying that both NSSI frequency and NSSI methods can help to distinguish high severity classes. 


The symposium will conclude with Ross Jacobucci (Jacobucci, Serang, & Ammerman) discussing the multivariate extensions of EDM, such as structural equation modeling trees, mixed effects decision trees, and regularized structural equation modeling, which combine exploratory and confirmatory aspects of statistical modeling. Specific applications to SITB research will be discussed.


 The presentations will be summarized and discussed by Matthew Nock. As one of the leading experts in suicidal and non-suicidal self-injury, he is well suited to provide thoughtful commentary on the research and clinical implications of these advances. 

Learning Objectives:

Taylor A. Burke

Doctoral Student in Clinical Psychology
Temple University

Presentation(s):

Send Email for Taylor Burke

Brooke A. Ammerman

Graduate Student
Temple University

Presentation(s):

Send Email for Brooke Ammerman

Send Email for Matthew Nock

Evan Kleiman

College Fellow
Harvard University

Presentation(s):

Send Email for Evan Kleiman

Taylor A. Burke

Doctoral Student in Clinical Psychology
Temple University

Presentation(s):

Send Email for Taylor Burke

Brooke A. Ammerman

Graduate Student
Temple University

Presentation(s):

Send Email for Brooke Ammerman

Ross Jacobucci

University of Notre Dame

Presentation(s):

Send Email for Ross Jacobucci


Assets

Symposium 35 - Exploratory Data Mining in Clinical Research: Demonstrating Applications in Self-Injurious Thoughts and Behaviors



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Exploratory Data Mining in Clinical Research: Demonstrating Applications in Self-Injurious Thoughts and Behaviors