Screening of depression and suicidality using voice biomarkers
Research type
Research Study
Full title
Development of a machine-learning algorithm for the screening and monitoring of depression and suicidality using voice biomarkers.
IRAS ID
353720
Contact name
Adrian Cree
Contact email
Sponsor organisation
Priory Group
Duration of Study in the UK
1 years, 0 months, 0 days
Research summary
Previous research has found that aspects of voice (e.g. volume, rhythm, harmony, etc.) may be linked to both mental and physical health conditions. Analysis of these voice biomarkers may create new opportunities for healthcare, allowing for quicker and more objective screening of health conditions.
The research team are collaborating with Peak profiling, a company who are developing the machine-learning algorithm to analyse patterns and biomarkers of our voice. The aim of the current study is to collect voice data from patients, within the Priory, which can be used to train the machine-learning algorithm for the detection of voice biomarkers related to depression. If the training of the voice data shows that a valid algorithm for depression (and suicidality) can be trained, a medical detection product will be developed in a later step, following the current research study, which would offer a non-invasive, remote and more objective detection of depression, and therefore will have many benefits for mental health services.
In collaboration with Priory, Peak Profiling have developed a research app, which participants will be asked to download and install to participate in the research. Participants will be asked to upload voice samples weekly, for a period of up to 5 weeks. Each voice sample will involve three vocal tasks, including counting from 1 to 10, describing a randomly assigned image and talking about one’s week for 60 seconds. At the time of participants completing first and last voice submissions, they will be asked to fill out the PHQ-9 and GAD-7. Participants will be asked about their socio-demographic data within the research app (age, gender and ethnicity). Patients’ ICD-10 diagnosis classification, current medication prescriptions and first three digits of postal codes will be obtained from a health records database. The vocal recordings will be used to train the algorithm to detect depression through voice biomarkers alone.REC name
London - Chelsea Research Ethics Committee
REC reference
25/LO/0570
Date of REC Opinion
5 Sep 2025
REC opinion
Further Information Favourable Opinion