A Novel Approach to Clustering Accelerometer Data for Application in Passive Predictions of Changes in Depression Severity
Authors:
- Mindy K. RossMRMindy K. Ross
- Theja TulabandhulaTTTheja Tulabandhula
- Casey C. BennettCBCasey C. Bennett
- EuGene BaekEBEuGene Baek
- Dohyeon KimDKDohyeon Kim
- Faraz HussainFHFaraz Hussain
- Alexander P. DemosADAlexander P. Demos
- Emma NingENEmma Ning
- Scott A. LangeneckerSLScott A. Langenecker
- Olusola AjiloreOAOlusola Ajilore
- Alex D. LeowALAlex D. Leow
Open Access
Abstract
The treatment of mood disorders, which can become a lifelong process, varies widely in efficacy between individuals. Most options to monitor mood rely on subjective self-reports and clinical visits, which can be burdensome and may not portray an accurate representation of what the individual is experiencing. A passive method to monitor mood could be a useful tool for those with these disorders. Some previously proposed models utilized sensors from smartphones and wearables, such as the accelerometer. This study examined a novel approach of processing accelerometer data collected from smartphones only while participants of the open-science branch of the BiAffect study were typing. The data were modeled by von Mises-Fisher distributions and weighted networks to identify clusters relating to different typing positions unique for each participant. Longitudinal features were derived from the clustered data and used in machine learning models to predict clinically relevant changes in depression from clinical and typing measures. Model accuracy was approximately 95%, with 97% area under the ROC curve (AUC). The accelerometer features outperformed the vast majority of clinical and typing features, which suggested that this new approach to analyzing accelerometer data could contribute towards unobtrusive detection of changes in depression severity without the need for clinical input.Keywords
Funding Information
- Mood Challenge for Research kit (1R01MH120168)
- National Institute of Mental Health (1R01MH120168)
This publication has 49 references indexed in Scilit:
- Clinical parameters of circadian rhythms in affective disordersEuropean Neuropsychopharmacology, 2011
- Predictors and moderators of time to remission of major depression with interpersonal psychotherapy and SSRI pharmacotherapyPsychological Medicine, 2010
- Retrospective recall of affect in clinically depressed individuals and controlsCognition and Emotion, 2009
- The cost of mood disordersPsychiatry, 2009
- Sleep in Mood DisordersSleep Medicine Clinics, 2008
- Diurnal mood variation in major depressive disorder.Emotion, 2006
- Monitoring Depression Treatment Outcomes With the Patient Health Questionnaire-9Medical Care, 2004
- The PHQ-9: A New Depression Diagnostic and Severity MeasurePsychiatric Annals, 2002
- Sleep and mood disordersSleep Medicine Reviews, 1997
- Algorithm AS 136: A K-Means Clustering AlgorithmJournal of the Royal Statistical Society Series C: Applied Statistics, 1979