The relationship between mobile phone location sensor data and depressive symptom severity

Abstract

Background: Smartphones offer the hope that depression can be detected using passively collected data from the phone sensors. The aim of this study was to replicate and extend previous work using geographic location (GPS) sensors to identify depressive symptom severity.

Methods: We used a dataset collected from 48 college students over a 10-week period, which included GPS phone sensor data and the Patient Health Questionnaire 9-item (PHQ-9) to evaluate depressive symptom severity at baseline and end-of-study. GPS features were calculated over the entire study, for weekdays and weekends, and in 2-week blocks.

Results: The results of this study replicated our previous findings that a number of GPS features, including location variance, entropy, and circadian movement, were significantly correlated with PHQ-9 scores (r's ranging from -0.43 to -0.46, p-values < .05). We also found that these relationships were stronger when GPS features were calculated from weekend, compared to weekday, data. Although the correlation between baseline PHQ-9 scores with 2-week GPS features diminished as we moved further from baseline, correlations with the end-of-study scores remained significant regardless of the time point used to calculate the features.

Discussion: Our findings were consistent with past research demonstrating that GPS features may be an important and reliable predictor of depressive symptom severity. The varying strength of these relationships on weekends and weekdays suggests the role of weekend/weekday as a moderating variable. The finding that GPS features predict depressive symptom severity up to 10 weeks prior to assessment suggests that GPS features may have the potential as early warning signals of depression.

Keywords: Depression; Depressive symptoms; Geographic locations; Mobile phone; Students.

Figure 1. PHQ-9 score distribution at baseline (A) and the end of study follow-up (C). (B) shows the change from baseline to follow-up.

Each line represents one participant.

Figure 2. Feature extraction procedure for 2-week features.

(A) The first set of features ( F₁ to F₉) were extracted from 2-week blocks of sensor data that had an overlap of 1 week. (B) The second set of features were extracted after each week of data was split into weekday (Monday to Friday) and weekend (Saturday and Sunday). Weekday features (F_WD,1 to F_WD,9) were extracted from the weekday part and weekend features (F_WE,1 to F_WE,9) from the weekend part of each 2-week block.

Figure 3. Mean temporal correlations between 2-week location features, calculated at different time points during the study, and baseline and follow-up PHQ-9 scores.

Error bars show the 95% confidence intervals. In (A–B), features were obtained from weekday data only, and in (C–D), they were extracted from weekend sensor data. For each 2-week feature set, week indices indicate when the 2-week period ended. Due to sparsity of data in week 10, we excluded it from this analysis.