Background: Services with an orientation to early intervention strive to intervene at the earliest opportunity in the onset of mental illness or relapse. However, clinicians remain at a disadvantage as they catch up with the mental state of their clients often days or weeks after a deterioration in a person’s mental state. mHealth provides the possibility of real time tracking of key psychophysiological indices of good mental health such as sleep, arousal and activity. This project is using a mHealth device to track and record these indices with the aim of recognizing signs of early deterioration in mental health.
Aims: Assess the effectiveness of integrating a mHealth device in the clinical care of young people with a mental illness.
Methods: A randomised controlled trial of the use of the E2 (empatica) device measuring sleep, arousal, and level of activity is being conducted in two youth mental health services. Data is being examined using machine learning to predict significant changes in mental state.
Results: Using a Random Forest Classifier we were able to predict clinical deterioration with a high degree of specificity (0.94) but a low sensitivity (0.33). This current project looks to improve upon this accuracy. Conclusion: We have been able to demonstrate the acceptability of wearing a mHealth device over an extended period. While reasonable accuracy was achieved it did not reach an acceptable level for clinical use. The presentation will report on the next iteration of this project in improving the usefulness of the data.