Currently accepted at: JMIR mHealth and uHealth
Date Submitted: Jun 1, 2026
Open Peer Review Period: Jun 4, 2026 - Jul 30, 2026
Date Accepted: Nov 17, 2026
(closed for review but you can still tweet)
This paper has been accepted and is currently in production.
It will appear shortly on 10.2196/103254
The final accepted version (not copyedited yet) is in this tab.
Privacy-Preserving Mobile Face Tracking for Assessing Empathic Facial Reactivity: Cross-Sectional Study in Healthy Adults
ABSTRACT
Background:
Psychiatric and neurological conditions are often accompanied by alterations in empathic resonance with emotional states of others, yet objective measures of affective empathy remain limited. Conventional behavioral assessments including physiological and computational methods contain practical limitations. Mobile and privacy-preserving facial expression analysis can address this measurement gap by integrating facial empathic reactivity into established empathy assessments.
Objective:
This study aimed to evaluate a novel iOS-based application for computationally efficient and privacy-preserving assessment of spontaneous facial reactivity in healthy adults; examining valence effects, temporal dynamics and convergence with subjective reports during multidimensional empathy paradigms with static and dynamic stimuli.
Methods:
The cross-sectional observational study recruited healthy adults through convenience sampling from the Degeneration Controls and Relatives (DANCER) cohort at the German Center for Neurodegenerative Diseases Bonn and targeted public advertising at University Hospital Bonn. Participants without current neuropsychiatric diagnosis or relevant sensory-motor impairment were included. Final sample comprised 49 participants for the Multifaceted Empathy Test-core-2 with static stimuli (22 women, 27 men; mean age 69.46, SD 8.51 years) and 40 completed an established task with film clips (mean age 70.16, SD 8.61 years). An iPad TrueDepth camera and Apple ARKit extracted 3D blend shape coefficients in real time without storing images. Camera activation was partially disclosed while spontaneous facial reactivity assessment was masked until full-debriefing. We examined 33 blend shape coefficients with linear mixed-effects models testing valence effects, temporal dynamics, task-demand effects and further demographic, state and trait empathy associations with required model adjustments. We conducted hierarchical clustering to explore coactivation patterns.
Results:
Positive versus negative condition contrasts elicited greater positive emotional expression activity such as mouthSmile in MET-core-2 (β=.199, 95% CI 0.094-0.304, P<.001) and film clips (β=.705, 95% CI 0.295-1.114, P=.003), alongside lower mouthFrown activity (MET-core-2: β=−.103, 95% CI −0.153 to −0.053, P<.001). This valence differentiation increased over time for most of the significant blend shapes. Reactivity towards negative stimuli was less uniform across features. Film clips elicited stronger positive-stimulus responses than static images such as for cheekSquint (β=.0310, 95% CI 0.0177-0.0444, P<.001). Task demands, age and sex variables selectively moderated facial responses across task parts. Clustering revealed recurring local coactivation patterns particularly cohesive in positive valence. Higher affective-empathy ratings were associated with a subset of blend shapes alongside mostly selective trait-empathy associations.
Conclusions:
The established mobile 3D face tracking captured subtle, spontaneous and temporally dynamic empathic facial reactivity without raw image storage. Extending expression research to established empathy tasks, this study provides convergent evidence for objective assessment of the facial component of empathic reactivity in healthy adults. The paradigm offers a practical method to complement self-report and performance measures in mobile research settings. Concurrent examinations with established measures and studies in clinical populations can improve the reliability and clinical utility of the setting.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.