Smartphone as a monitoring tool for bipolar disorder: a systematic review including data analysis, machine learning algorithms and predictive modelling.

Abstract

Bipolar disorder (BD) is a chronic illness with a high recurrence rate. Smartphones can be a useful tool for detecting prodromal symptoms of episode recurrence (through real-time monitoring) and providing options for early intervention between outpatient visits. The aim of this systematic review is to overview and discuss the studies on the smartphone-based systems that monitor or detect the phase change in BD. We also discuss the challenges concerning predictive modelling. Published studies were identified through searching the electronic databases. Predictive attributes reflecting illness activity were evaluated including data from patients' self-assessment ratings and objectively measured data collected via smartphone. Articles were reviewed according to PRISMA guidelines. Objective data automatically collected using smartphones (voice data from phone calls and smartphone-usage data reflecting social and physical activities) are valid markers of a mood state. The articles surveyed reported accuracies in the range of 67% to 97% in predicting mood status. Various machine learning approaches have been analyzed, however, there is no clear evidence about the superiority of any of the approach. The management of BD could be significantly improved by monitoring of illness activity via smartphone.