Geolocation-derived mobility indices and the association with clinical symptoms and functioning in severe mental illness: A multivariate meta-analysis.

Abstract

Geolocation-derived mobility indices from smartphone sensors may be a digital phenotype of negative symptoms in severe mental illness (SMI; schizophrenia and bipolar disorder). In this preliminary analysis, we compared SMI and healthy comparators (HC) on five geolocation indices (time at home, distances traveled, distances traveled from home, normalized entropy, and number of locations visited) and tested associations with clinical and functional outcomes. A systematic search identified eight eligible samples (k; n = 153 effects, nSMI = 291, nHC = 162 across all analyses). Effect sizes included Hedge's g for group differences and Pearson's r for correlations with symptoms and functioning. Multivariate mixed-effects models derived overall effect sizes. Compared to HC, significant reductions were observed in SMI for overall mobility (g = -0.45, k = 6), time at home (g = -0.62, k = 4), distances from home (g = -0.61, k = 2), and locations visited (g = -0.47, k = 5). Within SMI, overall mobility (r = -.19, k = 5), time at home (r = -.24, k = 3), and normalized entropy (r = -.29, k = 1) were associated with broadly defined negative symptom burden. Associations with functional outcomes were not significant (rs ≤ |.20|, ks ≤ 3). Mobility within activity space was moderately diminished in SMI. Indices reflecting restricted activity space and location variability were modestly associated with negative symptoms in SMI. Geolocation indices alone provide weak symptom monitoring information, although the paucity of studies limits the robustness and generalizability of these preliminary findings. Recommendations are discussed. (PsycInfo Database Record (c) 2025 APA, all rights reserved).