News

Article

Study Using Machine Learning Identifies Contributors to Social Isolation for Schizophrenia

Author(s):

Key Takeaways

  • Social anhedonia and nonsocial cognition are key predictors of social isolation in schizophrenia, explaining 14% and 21% of variance in isolation and loneliness, respectively.
  • LASSO regression models were used to evaluate interrelationships among social cognition, nonsocial cognition, depression, social anhedonia, and social avoidance motivation.
SHOW MORE

While this analysis found social anhedonia may be a potential intervention for some, nonsocial cognition may have a significant role in social disconnection for those with schizophrenia.

Study Using Machine Learning Identifies Contributors to Social Isolation for Schizophrenia

Samuel J. Abplanalp, PhD

Credit: ResearchGate

Social anhedonia contributes to social isolation and loneliness broadly, new machine learning findings suggest, though nonsocial cognition likely explains the unique variance seen in social isolation among those with schizophrenia.1

This new research was authored in part by Samuel J. Abplanalp, PhD, from the department of psychiatry and biobehavioral sciences at UCLA’s Jane and Terry Semel Institute for Neuroscience and Human Behavior.

Abplanalp et al. noted that regression-based machine learning models using a Least Absolute Shrinkage and Selection Operator (LASSO) are helpful for evaluating a wide range of potential interrelationships among variables.2 This was helpful given the plethora of factors contributing to social isolation and other mental health struggles.

“Therefore, we used LASSO regression to examine the degree to which social cognition, nonsocial cognition, depression, social anhedonia, and social avoidance motivation were linked to social isolation and loneliness in schizophrenia, a psychiatric comparison sample (BD), and a CS enriched for social isolation,” Abplanalp and colleagues wrote. “We evaluated the relationships among these variables that were present within samples and how the relationships differed between samples.”1

Background and Trial Design

All participants in this analysis were a part of a larger research project assessing the psychological elements contributing to social disconnection. There were 72 outpatients who were involved in this study with schizophrenia, 48 who had bipolar disorder, and 151 who were drawn from the general community.

The investigators recruited clinical samples from several different outpatient clinics, 1 example of which was the Veterans Affairs Greater Los Angeles Healthcare System. They made diagnoses through the use of the Structured Clinical Interview for DSM-5 (SCID-518). When they were shown to be available, medical records were implemented for confirmation.

Study subjects with schizophrenia and bipolar disorder were all clinically stable, given a lack of hospitalizations in the prior 3 months and a lack of recent shifts in psychoactive medications within the prior 4 weeks. All such patients were actively being given psychoactive medications over the course of the assessment.

For the purposes of recruiting individuals who experienced high social isolation levels, the research team used advertisements which inquired whether readers had minimal family contact, few friends, and a lack of engagement in activities alone. Through such ads, the team enrolled 96 participants.

They also implemented online ads similar to those used in previous studies designed to find healthy controls, without specific mention of social isolation. The team received 55 responses from participants and were, thus, able to find their community sample including both socially isolated and non-isolated individuals. It was, however, skewed toward isolation.

Summary of Findings

In this study, using the aforementioned machine learning techniques to find predictors of social isolation as well as loneliness within the schizophrenia cohort (N = 72), the investigators looked at such factors. For the purposes of comparison, they also included a group of subjects with bipolar disorder (N = 48) and a community sample (N = 151).

Similar social isolation levels were seen among all study subjects, and the research team found that there had not been statistically significant differences among them [F (2,268) = 1.59, P = .213, η2 = .012]. Within the community arm of the study, the team reported that 27 met the criteria for a personality disorder, including diagnoses such as borderline, avoidant, schizoid, paranoid, and schizotypal disorders.

Among those in the schizophrenia arm, the investigators’ model predicting social isolation explained 14% of the observed variance (R2 = .14), while the team’s model for loneliness was shown to have explained 21% (R2 = .21). The investigators added that social anhedonia (β = .14, 100% of runs) and nonsocial cognition (β = −.09, 96.66% of runs) were found to be the only significant social isolation predictors for the schizophrenia sample.

“Social isolation and loneliness are associated with substantial public health risks,” they wrote. “The present study used machine learning to identify social anhedonia as a transdiagnostic variable that is linked to social isolation and loneliness in schizophrenia, BD, and a CS enriched for social isolation.”1

References

  1. Abplanalp SJ, Green MF, Wynn JK, et al. Using machine learning to understand social isolation and loneliness in schizophrenia, bipolar disorder, and the community. Schizophr 10, 88 (2024). https://doi.org/10.1038/s41537-024-00511-y.
  2. Signorino CS & Kirchner A. Using LASSO to model interactions and nonlinearities in survey data. Surv. Pract. 11, 1–0 (2018).
Related Videos
Caroline Piatek, MD: Improving Patient-Reported Outcomes in PNH With Danicopan Add-on Therapy
Haydar Frangoul, MD: Preventing VOCs in People With Sickle Cell Disease With Exa-Cel Gene Editing Therapy
Andreas Kremer, MD, PhD, MHBA | Credit: AASLD
Andreas Kremer, MD, PhD, MHBA | Credit: AASLD
Andreas Kremer, MD, PhD, MHBA | Credit: AASLD
Andreas Kremer, MD, PhD, MHBA | Credit: AASLD
Jörn Schattenberg, MD | Credit: Novo Nordisk
Diabetes Dialogue: Diabetes Tech Updates from November 2024 | Image Credit: HCPLive
© 2024 MJH Life Sciences

All rights reserved.