Smartphone Keystroke Biomarkers as Predictors of Adverse Neuropsychiatric Sequelae After Trauma in Trauma Survivors: Prospective Observational Cohort Study.

BACKGROUND

Adverse posttraumatic neuropsychiatric sequelae are common after trauma. Early identification of individuals at risk for these outcomes could enable the deployment of preventive interventions to survivors at greatest risk. Smartphone keystroke biomarkers show promise in identifying individuals with neuropsychiatric symptoms; however, to our knowledge, no research has examined whether they can be used to identify symptoms in the aftermath of trauma.

OBJECTIVE

This study evaluates whether passively collected keystroke data from smartphone use in daily life could identify individuals with high symptom levels, as well as worsening or recovery of symptoms, after trauma exposure.

METHODS

Data from a diverse cohort of individuals presenting to 27 emergency departments after trauma were analyzed. Inclusion criteria were presenting to the emergency department within 72 hours of trauma, age 18-75, and the ability to speak and read English. Exclusion criteria were solid organ injury, significant hemorrhage, operative intervention, or likely admission for over 72 hours. Participants installed an app that passively collected keystroke data during use of any app on their smartphone, beginning in the emergency department. Participants also completed serial symptom assessments over 8 weeks after trauma exposure.

RESULTS

A total of 3445 patients met study criteria, provided informed consent, and completed assessments in the emergency department. Of these, 1072 (mean age 40, SD 13; 616/1072, 57.46%, women; 565/1072, 52.71% non-Hispanic Black) installed the app on their Android smartphone and completed the 8-week assessment and were therefore included in analyses. Keystroke biomarkers related to typing speed, identified using bivariate linear mixed models controlling for false discovery rates, were associated with elevated pain, reexperiencing, and mental fatigue (absolute values of rs=0.22-0.25, Ps=.02). Separate change-of-operation and scrolling keystroke biomarkers were associated with increased reexperiencing symptoms (r=0.18, P=.047) and mental fatigue (rs=0.18-0.19, Ps=.031-.047). Further, changes in specific keystroke biomarkers were associated with worsening or recovery of pain (rs=0.07-0.10, Ps=.02), somatic symptoms (rs=0.02, Ps=.02), mental fatigue (rs=0.02-0.04, Ps=.02), sleep disturbance (absolute rs=0.07-0.09, Ps=.02), reexperiencing (rs=0.02-0.04, Ps=.02), and hyperarousal (rs=0.02-0.04, Ps=.02).

CONCLUSIONS

In general, slower typing and scrolling speeds were associated with higher symptom levels, with small to medium effect sizes. Keystroke data passively collected via smartphone use may help identify individuals with significant or changing posttraumatic symptoms. Future research should continue to explore these keystroke biomarkers and whether they can be leveraged to connect vulnerable trauma survivors to appropriate services. Overall, these results add to the literature, indicating that passively collected keystroke data may help identify individuals with neuropsychiatric symptoms or changes and are, to our knowledge, the first to test whether keystroke biomarkers are useful in the aftermath of trauma. This represents a critical period during which preventive interventions could be deployed to reduce the long-term burden of trauma-related sequelae.