Passive Smartphone Sensors for Detecting Psychopathology

Area of Science:

• Digital Phenotyping
• Computational Psychiatry
• Behavioral Science

Background:

• Smartphone sensors offer continuous, unobtrusive data collection for clinical insights.
• Progress in identifying unique behavioral markers for psychopathology is hindered by heterogeneous diagnostic categories.

Purpose of the Study:

• To determine which psychopathology domains are detectable via smartphone sensors.
• To identify passively sensed markers for general impairment (p-factor) and specific transdiagnostic domains.

Main Methods:

• Cross-sectional study utilizing data from the Intensive Longitudinal Investigation of Alternative Diagnostic Dimensions.
• Collected baseline surveys and 15 days of smartphone monitoring (GPS, accelerometer, call logs, screen status, battery).
• Analyzed 27 behavioral markers against transdiagnostic psychopathology dimensions (internalizing, detachment, disinhibition, antagonism, thought disorder, somatoform) and the p-factor.

Main Results:

• Detachment, somatoform, and internalizing showed the strongest correlations with sensed behaviors.
• 52% of sensor variables (14 of 27) were associated with psychopathology domains.
• Unique behavioral signatures were identified for most domains, with the p-factor linked to reduced mobility, more time at home, later bedtimes, and lower phone charge.

Conclusions:

• Tethering transdiagnostic domains to behavioral markers maximizes mobile sensing potential for studying psychopathology mechanisms.
• Findings can inform the development of symptom monitoring tools to improve clinical practice.
• Future research can translate these insights into more precise and effective treatments for mental health conditions.