Skip to main content
conference

Predicting Schizophrenia States with Behavioral Sensing and Network Analysis

Pervasive Computing Technologies for Healthcare (Pervasive Health 2025), 2026

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. Related photograph from Individual behavioral insights in schizophrenia: A network analysis and mobile sensing approach. Credit: Image courtesy of Dr. Omar Costilla Reyes.
Figure 1 shared with the team’s June 2023 announcement: one participant’s reported symptoms decrease from November 2015 and then gradually improve, and the authors focus on the clinical interventions that precede such decreases. Related photograph from Counterfactual explanations and predictive models to enhance clinical decision-making in schizophrenia using digital phenotyping. Credit: Image courtesy of Dr. Omar Costilla Reyes.

Research summary

A PervasiveHealth 2025 paper asks whether continuous behavioral sensing can predict individual psychological-network measures built from self-reports. The data come from people with schizophrenia.

The paper and the question

By Andy Davies, Hane Aung, and Omar Costilla-Reyes, the paper builds on the authors’ earlier individual-level study of schizophrenia. In that study, repeated ecological momentary assessments (EMA), brief in-the-moment self-reports collected on a phone, were turned into a psychological network: a map of how a person’s reported experiences are statistically related. Here the authors ask whether behavioral-sensing data recorded continuously by the phone can predict those network measures, rather than only summary scores of the self-reports.

Methods and findings

The authors use Random Forest regression, a method that averages the predictions of many decision trees, and compare population-level and personalized models with predictions based on aggregated EMA scores. The publisher abstract reports that network measures were predicted more accurately, especially with network regularization and personalized models. The result suggests that relationships among a person’s reported experiences may add information alongside summary scores. The study describes a data-analysis method that treats a person’s reported experiences as a network, pointing toward individualized monitoring.

Conference program and related work

The EAI PervasiveHealth 2025 final program lists the paper in a paper session on mental health sensing and just-in-time support at the conference held in Eindhoven, the Netherlands, in October 2025. Springer Nature lists the paper’s online publication date as July 2026.

The earlier study, “Individual behavioral insights in schizophrenia,” appeared at PervasiveHealth 2023; both belong to Dr. Costilla Reyes’s digital phenotyping research, developed with Davies and Aung of the University of East Anglia.

Publication details

Authors
Andy Davies, Hane Aung, Omar Costilla-Reyes
Venue
Pervasive Computing Technologies for Healthcare (Pervasive Health 2025)
Year
2026
Pages
197-211
Publisher
Springer Nature Switzerland

BibTeX citation

@inproceedings{costilla2026predictingschizophreniastates,
  title = {Predicting Schizophrenia States with Behavioral Sensing and Network Analysis},
  author = {Andy Davies and Hane Aung and Omar Costilla-Reyes},
  year = {2026},
  booktitle = {Pervasive Computing Technologies for Healthcare (Pervasive Health 2025)},
  doi = {10.1007/978-3-032-27582-0_13},
}

Topics