Mobile Phone Sensor Correlates of Depressive Symptom Severity in Daily-Life Behavior: An Exploratory Study

Abstract

Background: Depression is a common, burdensome, often recurring mental health disorder that frequently goes undetected and untreated. Mobile phones are ubiquitous and have an increasingly large complement of sensors that can potentially be useful in monitoring behavioral patterns that might be indicative of depressive symptoms.

Objective: The objective of this study was to explore the detection of daily-life behavioral markers using mobile phone global positioning systems (GPS) and usage sensors, and their use in identifying depressive symptom severity.

Methods: A total of 40 adult participants were recruited from the general community to carry a mobile phone with a sensor data acquisition app (Purple Robot) for 2 weeks. Of these participants, 28 had sufficient sensor data received to conduct analysis. At the beginning of the 2-week period, participants completed a self-reported depression survey (PHQ-9). Behavioral features were developed and extracted from GPS location and phone usage data.

Results: A number of features from GPS data were related to depressive symptom severity, including circadian movement (regularity in 24-hour rhythm; r=-.63, P=.005), normalized entropy (mobility between favorite locations; r=-.58, P=.012), and location variance (GPS mobility independent of location; r=-.58, P=.012). Phone usage features, usage duration, and usage frequency were also correlated (r=.54, P=.011, and r=.52, P=.015, respectively). Using the normalized entropy feature and a classifier that distinguished participants with depressive symptoms (PHQ-9 score ≥5) from those without (PHQ-9 score <5), we achieved an accuracy of 86.5%. Furthermore, a regression model that used the same feature to estimate the participants' PHQ-9 scores obtained an average error of 23.5%.

Conclusions: Features extracted from mobile phone sensor data, including GPS and phone usage, provided behavioral markers that were strongly related to depressive symptom severity. While these findings must be replicated in a larger study among participants with confirmed clinical symptoms, they suggest that phone sensors offer numerous clinical opportunities, including continuous monitoring of at-risk populations with little patient burden and interventions that can provide just-in-time outreach.

Keywords: activities of daily living; classification; cluster analysis; depression; mobile health (mHealth).

Conflict of interest statement

Conflicts of Interest: None declared.

Figures

Figure 1

Example GPS location data, overlaid on satellite image. Each small circle represents a histogram bin, which has a size of 500 by 500 meters. The colors indicate the number of samples captured by each bin (brighter means more samples). The bigger blue circles show the center of the clusters detected by the clustering algorithm.

Figure 2

Example phone usage data from a participant. Each row is a day, and the black bars show the extent of time during which the phone has been is use. The bars on the right side show the overall phone usage duration for each day.

Figure 3

Scatter plots for location and phone usage data versus PHQ-9 scores, respectively. The coefficient of correlation between each feature and PHQ-9 scores and its corresponding P-value is shown on top of each plot. Solid and dashed lines, shown only for strong correlations (P<.05), show the fitted regression model and +/- root mean square deviation from the model, respectively.

Figure 4

Comparison of location and usage feature statistics between participants with no symptoms of depression (blue) and the ones with (red). Feature values are scaled between 0 and 1 for easier comparison. Boxes extend between 25th and 75th percentiles, and whiskers show the range. Horizontal solid lines inside the boxes are medians. One, two, and three asterisks show significant differences at P<.05, P<.01, and P<.001 levels, respectively (ENT, entropy; ENTN, normalized entropy; LV, location variance; HS, home stay; TT, transition time; TD, total distance; CM, circadian movement; NC, number of clusters; UF, usage frequency; UD, usage duration).

Figure 5

Coefficients of correlation between location features. One, two, and three asterisks indicate significant correlation levels at P<.05 p and respectively entropy entn normalized lv location variance hs home stay tt transition time td total distance cm circadian movement nc number of clusters>