Background: The incorporation of mobile sensor data in clinical care is becoming increasingly useful in studying a patient’s daily routine. We explore different approaches of measuring the changes in these routines by analyzing the Global Positioning System (GPS) sensor data from a participant’s mobile device. In particular, we investigate a variant of Dynamic Time Warping (DTW) to determine whether patients with high symptom severity have less dissimilarity in their routes.
Methods: Using GPS data of individuals with Schizophrenia, we examine the similarity in their daily spatiotemporal patterns. Moreover, we analyze 42 participants with Schizophrenia using the open-source application mindLAMP installed on the participants’ mobile devices to collect active and passive data. Participants completed the Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), and Positive and Negative Syndrome Scale (PANSS) surveys prior to data collection. After removing participants with insufficient GPS data, we have 28 participants for analysis. The average Dynamic Time Warping (DTW) distance was computed for each participant’s daily GPS trajectories throughout the study using an implementation of FastDTW.
Results: We observe a weak negative association between Mean DTW and PANSS Total score (r = -0.29, p-value = 0.1214). In addition, we observe weak negative association between Mean DTW and PANSS Negative Subscale score (r = -0.31, p-value = 0.0977). Given the negative association, this would imply that participants in the sample who had lower symptom severity prior to data collection proceeded to have greater dissimilarity in their GPS data on a daily basis.
Conclusions: In our sample of Schizophrenia participants, we observed a negative association between symptom severity and route dissimilarity.