Digital biomarkers of mood disorders and symptom change

Nicholas C Jacobson,Sabine Wilhelm,Hilary Weingarden

doi:10.1038/s41746-019-0078-0

Nicholas C Jacobson, Sabine Wilhelm + Show 1 more

Open Access

https://doi.org/10.1038/s41746-019-0078-0

Copy DOI

Abstract

Current approaches to psychiatric assessment are resource-intensive, requiring time-consuming evaluation by a trained clinician. Development of digital biomarkers holds promise for enabling scalable, time-sensitive, and cost-effective assessment of both psychiatric diagnosis and symptom change. The present study aimed to identify robust digital biomarkers of diagnostic status and changes in symptom severity over ~2 weeks, through re-analysis of public-use actigraphy data collected in patients with major depressive or bipolar disorder and healthy controls. Results suggest that participants’ diagnostic group status (i.e., mood disorder, control) can be predicted with a high degree of accuracy (predicted correctly 89% of the time, kappa = 0.773), using features extracted from actigraphy data alone. Results also suggest that actigraphy data can be used to predict symptom change across ~2 weeks (r = 0.782, p = 1.04e-05). Through inclusion of digital biomarkers in our statistical model, which are generalizable to new samples, the results may be replicated by other research groups in order to validate and extend this work.

Highlights

Mood disorders occur in 12% of the population in their lifetime, resulting in substantial functional and economic burden.[1]
Movement data from actigraphs may be especially useful for detecting MDD or bipolar disorders, because these disorders are characterized by notable increases or decreases in goal-directed behavior, energy level, and movement, in addition to disruption in sleep— behavioral shifts which are likely to be captured via actigraph
One investigation used daytime and nighttime movement data from actigraphs worn on wrists of patients with primary bipolar disorder or MDD and controls to classify participants’ diagnostic group, using support vector machines.[4]

Summary

INTRODUCTION

Mood disorders (i.e., major depressive disorder [MDD], bipolar I, bipolar II) occur in 12% of the population in their lifetime, resulting in substantial functional and economic burden.[1]. A small base of research has explored using actigraphy data for mood disorder diagnosis.[2,3] One investigation used daytime and nighttime movement data from actigraphs worn on wrists of patients with primary bipolar disorder or MDD and controls to classify participants’ diagnostic group (kappa = 0.443, accuracy = 72.7%), using support vector machines (a machine-learning algorithm, which attempts to create the greatest difference between groups on dimensional hyperplanes).[4] the authors utilized features that were context-dependent (data were directly referenced to time within study, e.g., movement on minute 1 within the study), rather than time-relative (e.g., the consistency of movement between 1 day to the on average within the sample) This limits the generalizability of their methods to new samples, because linking data to absolute time does not allow for generalization to shorter or larger timespans. We sought to use methods that are generalizable to new samples, so that results may be replicated by other research groups

RESULTS

DISCUSSION

Diagnostic Status

Participants