Skip to main content
conference

Individual behavioral insights in schizophrenia: A network analysis and mobile sensing approach

International Conference on Pervasive Computing Technologies for Healthcare, 2024

Methods graphic from the team’s December 2023 announcement: self-reports and passively recorded sensor measures are filtered by behavior and compared as networks in an n-of-1 analysis. Credit: Image courtesy of Dr. Omar Costilla Reyes.

Research summary

A study of people with schizophrenia combines smartphone sensing, brief self-reports (EMA), and individual-level network analysis. It examines how behavior and reported experience relate over time.

The study

Presented at the International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth 2023) and published by Springer in 2024, the paper by Andy Davies, Eiko I. Fried, Omar Costilla-Reyes, and Hane Aung brings together computer science, psychology, and AI from the University of East Anglia, Leiden University, and MIT. Using CrossCheck data from people with schizophrenia, the study builds individual networks from repeated ecological momentary assessments (EMA), brief in-the-moment self-reports collected on a phone, and relates them to behavioral context measured by the phone’s sensors, such as the number of locations visited in a day.

Networks of reported experience

In a network of this kind, each reported experience is a node and the statistical association between two experiences is a link. The analysis is deliberately conducted at the individual, or n-of-1, level rather than across the population: the authors argue that the diversity and complexity of psychiatric disorders make population-level black-box prediction the wrong question. The method combines network psychometrics, statistical analysis, and digital phenotyping.

Findings and connections

The researchers found that networks observed during changes in behaviors such as the number of locations visited differed significantly from baseline and randomly sampled periods; for example, staying home altered the structure of positive EMA responses. The finding shows that a person’s measured behavior and reported experience are linked in ways that individual networks can reveal. The authors present the work as the foundation for a principled method for consolidating subjective self-report with concurrent behavioral sensor data in mental-health research, with the longer-term aim of supporting personalized treatment.

The study belongs to Dr. Costilla Reyes’s digital phenotyping research, which also includes the 2023 preprint on counterfactual explanations in schizophrenia. A follow-up by Davies, Aung, and Costilla-Reyes, listed in the PervasiveHealth 2025 program, asks whether behavioral sensing can predict such network measures directly.

Publication details

Authors
Andy Davies, Eiko I. Fried, Omar Costilla-Reyes, Hane Aung
Venue
International Conference on Pervasive Computing Technologies for Healthcare
Year
2024
Pages
18-33
Publisher
Springer Nature Switzerland

BibTeX citation

@inproceedings{costilla2024individualbehavioralinsights,
  title = {Individual behavioral insights in schizophrenia: A network analysis and mobile sensing approach},
  author = {Andy Davies and Eiko I. Fried and Omar Costilla-Reyes and Hane Aung},
  year = {2024},
  booktitle = {International Conference on Pervasive Computing Technologies for Healthcare},
  doi = {10.1007/978-3-031-59717-6_2},
}

Topics