Digital Relapse Prediction

Digital Relapse Prediction

Project Overview

Schizophrenia and depression are among the most disabling disorders in all of medicine. The costs to society are greater than nearly any other chronic health condition and the burden to patients and family members is of the greatest magnitude. Better measurement of patients lived experiences is critical to better understanding not only the mechanisms of these illnesses but also new opportunities to intervene in the future. Our team, with the help of an approved vendor, has built a novel platform app (Learn, Assess, Manage and Prevent “LAMP”) that syncs with a smartphone and captures physiology data. The goal of this study is to utilize LAMP to identify digital biomarkers of mental illness that will help us learn about what factors influence both recovery as well as relapse in patients with schizophrenia and depression.

Evaluated with the groups Below

100 patients in active treatment for schizophrenia or depression, and 100 healthy controls


The LAMP study lasts 12 months and consists of three study visits. If participants sign informed consent, the LAMP app and the Beiwe app, a research tool that records anonymous phone and text logs as well as GPS data, are downloaded onto their smartphones and they are given a smart watch to wear.  They will also take a paper/pencil neuropsychiatric battery of tests on symptoms and cognition. Subjects use the apps and wear the smart watch for 1 year. The apps prompt symptom surveys and cognitive tests three times a week, each taking five minutes to complete. After 12 months, the subjects will return to the clinic and take a repeat neuropsychiatric battery.


We will use novel methods to collect from multiple data streams, in this case mobile survey assessments, mobile cognitive assessments, and behavioral and social data gathered through passive data streams. The data will be analyzed using the R programming language, and we intend to use lasso regression, which is a machine-learning model that produces clinically interpretable classification results on large and diverse data sets. This model is built into the R language, an increasingly common analysis method for big data.


This study aims to assess self-reported, behavioral, cognitive, and physiological data gathered from smartphones and smart watches as compared to gold standard measured in patients with depression or schizophrenia. To determine whether the digital biomarkers our team identified are unique to mental illness, we will compare our findings to a control group. Our findings will help to determine which digital biomarkers are associated with relapse, and inform relapse prevention in patients with schizophrenia and depression going forward.


Relevant Reading
  • Torous J, Firth J, Mueller N, Onnela JP,Baker JT. Methodology and Reporting of Mobile Heath and Smartphone Application Studies for Schizophrenia. Harvard review of psychiatry. 2017 May1;25(3):146-54.
  • Firth J, Torous J, Nicholas J, Carney R,Pratap A, Rosenbaum S, Sarris J. The efficacy of smartphone‐based mental health interventions for depressive symptoms: a meta‐analysis of randomized controlled trials. World Psychiatry. 2017 Oct;16(3):287-98.
  • Ben-Zeev D, Brian R, Wang R, Wang W, CampbellA T, Aung MS, Merrill M, Tseng VW, Choudhury T, Hauser M, Kane JM. Cross Check:Integrating Self-Report, Behavioral Sensing, and Smartphone Use to Identify Digital Indicators of Psychotic Relapse.
  • Naslund JA, Aschbrenner KA, Bartels SJ.Wearable devices and smartphones for activity tracking among people with serious mental illness. Mental health and physical activity. 2016 Mar31;10:10-7.

Featured Projects