Deep learning predicting mood changes in depression (LEARN-MOOD)

  • Research type

    Research Study

  • Full title

    Using deep learning to predict mood changes in patients with depression

  • IRAS ID

    279220

  • Contact name

    Valeria Mondelli

  • Contact email

    valeria.mondelli@kcl.ac.uk

  • Sponsor organisation

    King's College London

  • Duration of Study in the UK

    0 years, 11 months, 17 days

  • Research summary

    Heart rate dynamics have previously been shown to predict symptoms of depression and human emotional state. Limbic Limited have developed a deep learning algorithm which predicts human emotional state using physiological data collected from consumer wearable devices (a wristwatch), as well as supporting information such as age, weight and gender. The information collected via wristwatch fitness trackers is then sent to dedicated servers for analysis. The algorithms currently return predictions about the wearers emotional state. We now seek to advance and validate their technology in the mental health field by testing whether the algorithms can accurately predict emotional state in patients with major depressive disorder. As well as heart rate dynamics, biological markers of stress such as cortisol, and inflammatory markers such as, IL-1β, IL-6, tumor necrosis factor (TNF) and C-reactive protein (CRP), have also been shown to predict emotional state. As well as measuring heart rate dynamics in relation to emotional state, this study will also measure biological markers of inflammation and stress from blood and saliva to test whether integration of biological markers with Limbic's algorithm further increases the prediction of variation in mood.

    Due to COVID-19, it may be necessary to carry out this study remotely. Parts of this study's protocol has been altered to enable the study to go ahead remotely i.e. without any face-to-face contact with participants. The main changes will be outlined here. Please refer to the study protocol for more details.

    1) Assessments that would have been completed within the NIHR King's Wellcome Clinical Research Facility (CRF), will be completed virtually via a video or telephone call.
    2) The biological sampling (i.e. blood and saliva samples) will not be carried out.
    3) Rather than recruiting 50 patients, up to 70 patients will be recruited.
    4) Study equipment (wristwatch fitness tracker) will be posted to participants.

  • REC name

    London - Surrey Research Ethics Committee

  • REC reference

    20/LO/0709

  • Date of REC Opinion

    26 Jun 2020

  • REC opinion

    Further Information Favourable Opinion