Symposium 56 - Promising Paths and Potential Perils in Integrating Machine Learning and EMA in Clinical Science

Chair(s)

• 

  Natalia Van Doren, Ph.D. (she/her/hers)

  NIDA T32 Postdoctoral Research Fellow
  University of California, San Francisco
  Palo Alto, California, United States
• 

  Nur Hani Zainal, M.S., Ph.D. (she/her/hers)

  Assistant Professor
  National University of Singapore
  Kent Ridge, Singapore

Discussant(s)

• JC

  Joshua Curtiss, M.A., Ph.D.

  Assistant Professor in Applied Psychology
  Northeastern University
  Boston, Massachusetts, United States

Presenter(s)

• AC

  Adam Calderon, B.S., M.A.

  The Pennsylvania State University
  University Park, Pennsylvania, United States
• AS

  Alison Schreiber, Ph.D. (she/her/hers)

  University of Pittsburgh School of Medicine
  Pittsburgh, Pennsylvania, United States
• 

  Aaron Fisher, Ph.D.

  Associate Professor
  University of California, Berkeley
  Berkeley, California, United States
• SB

  Seung Yeon Baik, M.S. (she/her/hers)

  Doctoral Student
  The Pennsylvania State University
  University Park, Pennsylvania, United States

Machine learning (ML) techniques promise to transform mental healthcare via innovative solutions, but their use must mitigate healthcare disparities. A way to advance clinical science in this area is by using ML to assess culturally diverse samples and transdiagnostic processes. As an example, enhanced comprehension of the interconnections between insomnia and conditions such as anxiety, depressive disorders, disordered eating, and alcohol use is imperative, given its high occurrence among diverse young adults. Supervised ML may enhance precision in assessing the discriminative accuracy of insomnia-related mental disorders. When coupled with Bayesian network analysis, it may elucidate the directionality between symptoms and their correlations. The first presenter, thus, highlights how these novel techniques indicated that depressed mood, fatigue, and self-esteem were the primary symptoms associated with insomnia in an extensive, diverse sample (N = 31,285) of young adults.

Another transdiagnostic process ML may offer further insights into involves facial emotional processing, as social dysfunction is salient in various mental disorders. For instance, whereas certain studies indicated improved facial emotion recognition in borderline personality disorder (BPD), others show diminished performance or suggest that outcomes were contingent upon the situation. One drawback of these studies is the predominant reliance on summary statistics of task performance with poor reliability to describe changes in facial emotion processing. The second presenter, thus, shows how ML-based computational psychiatry methods address this concern by directly incorporating uncertainty into parameter estimation.

ML could also be integrated with ecological momentary assessments (EMAs) to boost ecological validity, reduce recall bias, and enhance reliability via repeated sampling to better understand avoidance patterns in daily life. For instance, many clients worry to avoid sharp rises in negative emotions and to increase the odds of experiencing sharp rises in positive emotions. Harnessing both EMA and supervised ML, the third presenter explores how worry and rumination influence positive emotional contrasts following relief and positive events.

Unifying ML and EMA could also provide clinical researchers with insights regarding the timing and conditions of specific clinically relevant events. Momentary binary data can serve the dual purpose of indicating the presence or absence of a specific clinical event while also enabling the evaluation of concurrent constructs and processes. Thus, the final presenter shows how Ising models – network models designed for binary variables – were employed to uncover and elucidate the causal sequences of events linking context to eliciting, responding, and regulating emotions. Case examples will illustrate how these logical sequences function akin to relay mechanisms — establishing if-then rules governing the succession of context, emotion, and behavior, potentially guiding personalized treatment.

Finally, a seasoned discussant will synthesize these talks and propose ideas to harness ML and EMA to advance clinical science and potentially improve patient care.

Presentations:

• 2:30 PM – 4:00 PM EST

  (SYM 56) Machine Learning and Bayesian Network Analyses Identifies Associations with Insomnia in a National Sample of 31,285 Treatment-seeking College Students
  

  Speaker: Adam Calderon, B.S., M.A. – The Pennsylvania State University

  Co-author: Seung Yeon Baik, M.S. (she/her/hers) – The Pennsylvania State University

  Co-author: Matthew Ng, B.S. – Nanyang Technological University

  Co-author: Ellen Fitzsimmons-Craft, Ph.D. (she/her/hers) – Washington University School of Medicine

  Co-author: Daniel Eisenberg, Ph.D. – University of California, Los Angeles

  Co-author: Denise Wilfley, Ph.D. – Washington University School Of Medicine in St. Louis

  Co-author: C Barr Taylor, M.D. – Stanford University School of Medicine

  Co-author: Michelle G. Newman, B.S., M.A., Ph.D. (she/her/hers) – The Pennsylvania State University
• 2:30 PM – 4:00 PM EST

  (SYM 56)Enhanced Emotion Processing Following Exposure to Fear Stimuli in Young Adults with BPD: Applications of Drift Diffusion Models and Bayesian Distributional Regression Models to Task Behavior
  

  Speaker: Alison M. Schreiber, Ph.D. (she/her/hers) – University of Pittsburgh School of Medicine

  Co-author: Nate Hall, M.S. (he/him/his) – University of North Carolina

  Co-author: Daniel Parr, B.A. (he/him/his) – Duke University

  Co-author: Michael Hallquist, PhD – University of North carolina at chapel hill
• 2:30 PM – 4:00 PM EST

  (SYM 56) Deriving Clinical Inferences from Binary Momentary Data
  

  Speaker: Aaron J. Fisher, Ph.D. – University of California, Berkeley

  Co-author: Aaron J. Fisher, Ph.D. – University of California, Berkeley
• 2:30 PM – 4:00 PM EST

  (SYM 56) Why Do Individuals with Generalized Anxiety Disorder and Depression Engage in Worry and Rumination? : A Momentary Assessment Study of Positive Contrast Enhancement
  

  Speaker: Seung Yeon Baik, M.S. (she/her/hers) – The Pennsylvania State University

  Co-author: Seung Yeon Baik, M.S. (she/her/hers) – The Pennsylvania State University

  Co-author: Michelle G. Newman, B.S., M.A., Ph.D. (she/her/hers) – The Pennsylvania State University