JMIR Publications

ABSTRACT

Background:

Schizophrenia is a chronic condition, but the severity of symptomatic experiences and functional impairments vacillate over the course of illness. Developing unobtrusive remote monitoring systems to detect early warning signs of impending symptomatic relapses would allow clinicians to intervene prior to patients’ condition worsening.

Objective:

In this study, we created the first models, exclusively using passive sensing data from a smartphone, to predict behavioral anomalies that could indicate early warning signs of psychotic relapse.

Methods:

Data used to train and test the models were collected during the CrossCheck study. Hourly features derived from smartphone passive sensing data were extracted from 60 patients with schizophrenia (42 non-relapse, 18 relapse > 1 time throughout the study) and used to train models and test performance. We trained two types of encoder-decoder neural network models and a clustering-based local outlier factor model to predict behavioral anomalies that occur within the 30 day period prior to a participant's date of relapse. Models were trained to recreate healthy participant behavior, and then we applied a threshold to the recreation error to predict anomalies. The neural network model architecture was varied, as well as the percentage of relapse participant data used to train all models.

Results:

A total of 20,137 days of collected data was analyzed, with 726 days of data (0.037%) within any 30 day near-relapse period. The best performing model utilized a fully connected neural network autoencoder architecture with 80% of healthy days from relapse participant data for model training and 40 hidden units. The model achieved a median sensitivity of 0.25 and specificity of 0.88 (a 108% increase in behavioral anomalies near-relapse). We conducted a post-hoc analysis using the best performing model to identify behavioral features that had a medium to large effect (Cohen d≥0.5) in distinguishing behavioral anomalies prior to relapse from healthy days within four participants who relapsed multiple times throughout the study. Qualitative validation using clinical notes collected during the original CrossCheck study showed that the identified features from our analysis were presented to clinicians during relapse events.

Conclusions:

Our proposed method predicts a higher rate of behavioral anomalies in schizophrenia patients within the 30 day period prior to relapse, and can be used to uncover individual-level behaviors that change prior to relapse. This approach will enable technologists and clinicians to build unobtrusive digital mental health tools that can predict incipient relapse in schizophrenia spectrum disorders.