Mental Health World Conference 2026

Speakers - mhwc2026

Chan Yeong Cho Speaker at Mental Health conferences Summit

Chan Yeong Cho

Chan Yeong Cho

  • Designation: Departments of Psychiatry Chonnam National University Medical School, Republic of Korea
  • Country: Korea, Republic of
  • Title: Real World Behavioral Patterns in Schizophrenia Smartphone Derived Variability and Rainfall Reactivity

Abstract

Background
Previous smartphone-based studies in schizophrenia have primarily characterized average behavioral levels. This study examined temporal variability and rainfall-linked behavioral reactivity in patients with schizophrenia and controls. 
Study Design
We conducted a cross-sectional analysis of prospectively collected longitudinal GPS and step-count data from 87 patients with schizophrenia and 117 controls. Each of nine behavioral measures was summarized by its mean, standard deviation, coefficient of variation, and temporal slope, yielding 36 features. Seven classifiers were evaluated using participant level five old cros validation. Among participants with at least 10 valid days of matched behavioral and meteorological data, rainfall reactivity was calculated as the mean absolute within person Spearman correlation between daily precipitation and four behavioral measures
Study Results:
The mean only model achieved an AUROC of 0.738, compared with 0.858 when coefficients of variation and temporal slopes were included. The full Random Forest model achieved an AUROC of 0.860 and an AUPRC of 0.830 SHAP analyses identified the coefficients of variation for daily total steps and purposeful step ratio as the largest contributors to predictions. Among weather-matched participants, patients had lower weather reactivity (median, 0.096 vs. 0.153 P0.001) and fewer steps on rainy and non-rainy days (both P 0.001). Adding weather and activity reactivity features was associated with numerically higher performance (AUROC, 0.888 vs. 0.860 AUPRC, 0.850 vs. 0.830). In patients, weather reactivity was not significantly associated with PANSS N2, PANSS N4, or their combined score. 
Conclusions
These findings suggest that schizophrenia-related behavioral differences may be characterized by temporal variability and rainfall-linked modulation in addition to lower activity levels. Passive smartphone sensing may complement clinical assessment by characterizing behavioral organization in everyday life. However, the present findings do not establish it as a stand alone diagnostic tool.