(PS9-32) Smartphone Sensing in Suicide Research: Does Continuous Phone Monitoring Alter Phone Use and Provide Additional Information Beyond EMA?

Abstract Text: Background
Smartphone-based passive sensing has emerged as a promising tool for real-time suicide risk monitoring, potentially complementing active self-reporting. However, it remains unclear whether participants modify their behavior in response to monitoring; whether passive data is available when ecological momentary assessment (EMA) responses are missing; and whether information gained via passive data varies by study day or EMA compliance. To answer these questions, this study examines (1) whether screen time differs based on the presence of EMA prompts (e.g., increased awareness of monitoring), (2) whether screen time is captured on EMA-noncompliant days, and (3) whether its availability varies by day or EMA compliance.

Methods
Participants were 79 individuals with past-month suicidal thoughts (n = 79; 83% White; 35 ± 11 years; 75% women). Over a 35-day period, screen time was passively measured from participants’ smartphone use via screenshots captured every 5-seconds of active use. Participants also completed EMA surveys six times per day for 28 days, with half completing EMA in the first four weeks and the other half completing EMA in the second to fifth week. To examine whether monitoring awareness is associated with screen time, we compared the number of screenshots both between participants (i.e., those who started EMA immediately vs. those who started in the second week) and within participants (i.e., days with vs. without EMA prompts). We also assessed screen time on EMA-noncompliant days and tested whether its availability varied by study day or participants’ overall EMA compliance.

Results
Daily screen time in the first week did not differ between participants who received EMA immediately versus those who started in the second week (p = .22). Screen time did not differ within-person across weeks with and without EMA prompts (p = .81). Among days that participants did not respond to any EMA prompts, 12.31% had screen time captured. The likelihood of screen time capture on EMA-noncompliant days did not change over day (p = .45). However, participants with higher EMA compliance were more likely to have screen time data available on EMA-noncompliant days (b = 0.82, p = .04).

Conclusions
Results suggest awareness of phone monitoring did not alter screen time. Screen time was captured on some EMA-missing days, indicating that passive sensing can capture data beyond self-report. The likelihood of screen time being captured on these days remained stable over time and was higher among participants with greater EMA compliance, suggesting that data availability is influenced by individual engagement. These findings highlight passive sensing as a low-burden complement to EMA, providing continuous data even when self-reports are unavailable.