Abstract
BACKGROUND: Mood and anxiety disorders are highly prevalent and comorbid worldwide, with variability in symptom severity that fluctuates over time. Digital phenotyping, a growing field that aims to characterize clinical, cognitive and behavioral features via personal digital devices, enables continuous quantification of symptom severity in the real world, and in real-time. METHODS: In this study, N=114 individuals with a mood or anxiety disorder (MA) or healthy controls (HC) were enrolled and completed 30-days of ecological momentary assessments (EMA) of symptom severity. Novel real-world measures of anxiety, distress and depression were developed based on the established Mood and Anxiety Symptom Questionnaire (MASQ). The full MASQ was also completed in the laboratory (in-lab). Additional EMA measures related to extrinsic and intrinsic motivation, and passive activity data were also collected over the same 30-days. Mixed-effects models adjusting for time and individual tested the association between real-world symptom severity EMA and the corresponding full MASQ sub-scores. A graph theory neural network model (D(EP)NA) was applied to all data to estimate symptom interactions. RESULTS: There was overall good adherence over 30-days (MA=69.5%, HC=71.2% completion), with no group difference (t((58))=0.874, p=0.386). Real-world measures of anxiety/distress/depression were associated with their corresponding MASQ measure within the MA group (t's > 2.33, p's < 0.024). Physical activity (steps) was negatively associated with real-world distress and depression (IRRs > 0.93, p's ≤ 0.05). Both intrinsic and extrinsic motivation were negatively associated with real-world distress/depression (IRR's > 0.82, p's < 0.001). D(EP)NA revealed that both extrinsic and intrinsic motivation significantly influenced other symptom severity measures to a greater extent in the MA group compared to the HC group (extrinsic/intrinsic motivation: t((46)) = 2.62, p < 0.02, q FDR < 0.05, Cohen's d = 0.76; t((46)) = 2.69, p < 0.01, q FDR < 0.05, Cohen's d = 0.78 respectively), and that intrinsic motivation significantly influenced steps (t((46)) = 3.24, p < 0.003, q FDR < 0.05, Cohen's d = 0.94). CONCLUSIONS: Novel real-world measures of anxiety, distress and depression significantly related to their corresponding established in-lab measures of these symptom domains in individuals with mood and anxiety disorders. Novel, exploratory measures of extrinsic and intrinsic motivation also significantly related to real-world mood and anxiety symptoms and had the greatest influencing degree on patients' overall symptom profile. This suggests that measures of cognitive constructs related to drive and activity may be useful in characterizing phenotypes in the real-world.