Inferring Sleep Disturbance from Text Messages of Suicide Attempt Survivors: A Pilot Study
Ladis, Ilana, Psychology - Graduate School of Arts and Sciences, University of Virginia
Teachman, Bethany, AS-Psychology, University of Virginia
Sleep disturbance is a modifiable, acute risk factor for suicide, but naturalistic assessment of sleep problems is challenging. Examining communication on digital platforms can help identify phenotypes of sleep disturbance, which may aid in the detection of possible imminent suicide risk. This within-person pilot study examined sleep-related communication and texting patterns in a database of personal text messages (N = 86,705) provided by 26 individuals who survived at least one lifetime suicide attempt. Participants provided the dates of past suicide attempts, as well as two-week periods in which they experienced positive mood, depressed mood, or suicidal ideation. Generalized mixed effect models were used to test the association between suicide/mood episode type (e.g., attempt versus ideation) and three outcomes: likelihood of a text including sleep-related content or not, nightly count of texts sent from midnight-5:00 AM, and sum of unique hour bins from midnight - 5:00 AM with any outgoing text. Linguistic Inquiry Word Count was used to identify sleep-related texts, based on a custom dictionary of sleep-related words, phrases, and emojis. Analyses with a sleep dictionary that was manually revised to be more accurate showed that sleep-related communication was more likely during depressed mood episodes than positive mood episodes. Otherwise, there were no significant differences in either the likelihood of sleep-related communication, count of outgoing text messages from midnight – 5:00 AM, or sum of unique hour bins from midnight – 5:00 AM across suicide/mood episode types. Although sleep-related communication may differ as a function of within-person mood level, the present study did not detect differences in sleep-related communication tied to suicidal thoughts or behaviors. Future research with larger datasets and multiple data streams (e.g., call and social media logs) may provide insight into digital communication phenotypes associated with sleep problems and suicidal thoughts and behaviors.
MA (Master of Arts)
suicide, sleep, technology, risk-assessment, digital phenotyping
All rights reserved (no additional license for public reuse)