(PS10-83) RQA Reveals Associations Between the Dynamics of Negative Affective States and Symptoms of Depression and Anxiety

Background: Ecological momentary assessments (EMA) provide a more comprehensive and effective method of capturing real time measures of mood, activity, and mental health status. Although prior research has examined general temporal relationships across time-series measurements of affect and symptoms, little research has been pursued to uncover dynamic patterns in the structure of time-series data. In this study, recurrence quantification analysis (RQA) was used to examine patterns in the EMA responses of patients with emotional disorders across a two-week measurement window. RQA captures dynamic information within time-series data and can help highlight reoccurring patterns over time. These measures provide additional metrics by which to compare subjects and better understand the dynamics of their clinical presentation.

 

Methods: 33 Subjects completed a 2-week EMA protocol during which measures of specific mood states (Anxious, Cheerful, Content, Sad, and Upset) were assessed via smartphone 5 times per day. At the beginning and end of the two weeks, participants completed the Beck Depression Inventory-II (BDI-II) and the State Trait Anxiety Inventory (STAI). Four subjects were removed before analysis due to high degrees of missingness in their EMA responses ( > 40%), leaving a total of 29 subjects.  RQA measures of percent recurrence (%REC = total number of repeating values/total number of measurements) and percent determinism (%DET = total number of repeating patterns/total number of measurements) were calculated for each of the five moods states. Correlations between these affect specific RQA measures and the BDI-II and STAI were examined to determine whether dynamic recurrence metrics are associated with clinical symptom outcomes.

 

Results: Both %REC (Anxious: mean=0.26, s.d.=0.06; Cheerful: mean=0.29, s.d.=0.13; Content: mean=0.29, s.d.=0.11; Sad: mean=0.38, s.d.=0.17; Upset: mean=0.47, s.d.=0.23) and %DET (Anxious: mean=0.49, s.d.=0.12; Cheerful: mean=0.51, s.d.=0.15; Content: mean=0.50, s.d.=0.13; Sad: mean=0.61, s.d.=0.18; Upset: mean=0.72, s.d.=0.19) demonstrated high percentage values and variability across affect states, suggestive of potentially meaningful individual differences. Correlation analyses revealed significant negative correlations between post-EMA BDI-II scores and Sad %REC (r=-0.48, p=0.009) and %DET (r=-0.43, p=0.02), and Upset %REC (r=-0.53, p=0.002) and %DET (r=-0.50, p=0.005). Post-EMA STAI scores were significantly negatively correlated with Upset %REC (r=-0.40, p=0.03) and %DET (r=-0.38, p=0.04). Overall, BDI-II and STAI scores at week 2 demonstrated larger correlation coefficients with RQA measures of negative affective states compared to positive states.

 

Discussion: The results of this analysis indicate that EMA data and corresponding RQA values of %REC and %DET can provide meaningful and significant information about idiographic dynamics in affective states. Higher levels of %REC and %DET correspond to more structured and stable patterns within time-series. Thus, individuals who had more stable responses on measures of negative affect showed fewer symptoms of depression and anxiety as opposed to subjects with higher degrees of variability in their responses.