JMIR Publications

ABSTRACT

Background:

Mental health disorders are increasingly understood as dynamic systems in which symptoms interact over time. Daily smartphone surveys enable high-resolution tracking of daily-life experiences, while longitudinal network modeling allows to model symptom interactions over time. Individuals at clinical high risk for psychosis (CHR) frequently experience impaired cognition, functioning, and affect regulation alongside attenuated psychotic symptoms. Cross-sectionally, poorer cognition has been associated with lower functioning and resilience. However, cross-sectional assessments cannot capture the temporal and interdependent nature of these processes.

Objective:

In this study we aimed to characterize daily-life symptom dynamics in CHR and to examine how cognitive performance moderates network organization and dynamics. This is the first longitudinal network study in CHR that includes positive symptoms, affect, cognition and functioning.

Methods:

We used daily diary data collected over 60 days from 263 participants from the Accelerating Medicines Partnership® Schizophrenia Program. Baseline neurocognitive performance was assessed by the Penn Computerized Neurocognitive Battery. We used multilevel vector autoregressive models to estimate within-person networks capturing temporal and contemporaneous associations, and between-person networks reflecting individual differences. Individuals were classified into high and low cognition groups to assess differences in temporal and contemporaneous symptom networks. Furthermore, we examined how baseline cognitive domains relate to network characteristics, such as autocorrelations, network density and number of cycles present in the network.

Results:

In the temporal network positive affect showed the highest outgoing connection strength, temporally influencing next-day concentration, functioning, anhedonia, and positive symptoms (β=-0.07-0.08, P<0.013). In the contemporaneous network positive affect was again highly interconnected and demonstrated the strongest connections to all other variables. At the between-person level anhedonia emerged as the most central symptom. Individuals with higher baseline cognitive performance displayed stronger autocorrelations in social behavior and anhedonia (Δ=0.05-0.06, P=.024-.047). They also demonstrated a different directional association between positive symptoms and social behavior (Δ=-0.05, P=.024): fewer positive symptoms were linked to more social behavior, while the opposite was true for those with lower cognitive performance. Slower sensorimotor speed was associated with higher density of the temporal network (r=.22, PFDR=.044).

Conclusions:

Our findings demonstrate the central role of positive affect in within-person symptom dynamics in CHR. Our results furthermore indicate that baseline cognitive functioning does shape symptom network characteristics. In adaptive states, such as higher social engagement, cognitive functioning may support beneficial stability. Overall, these results emphasize the benefit of combining daily diary data with longitudinal network modeling to identify potential targets for digital interventions.