'
Jennifer Highland

Using voice analysis to identify peaks on bipolar disorder

Using voice analysis to identify peaks on bipolar disorder is a summary of a recently-published research paper titled “Voice analysis as an objective state marker in bipolar disorder.” For additional information, please click on the link.

The human voice is composed of multiple different components, created through complex muscle movements, making each person’s voice individual, like ‘a fingerprint’. Studies analysing speech in affective disorders, date back as early as 1938. Several clinical observations suggest that changes in speech features have been suggested as valid measures, to identify periods of depression and mania in bipolar disorder. For instance, reduced speech activity may be considered a symptom of depression, and increased speech activity may predict a switch to hypomania.

A pilot study conducted by The Copenhagen Clinic for Affective Disorders, with patients with bipolar disorder, aimed to investigate the following:

  1. Voice features collected during phone calls, as objective markers of affective states in bipolar disorder
  2. If combining voice features with automatically generated objective smartphone data on behavioural activities (for example, number of text messages and phone calls per day) and electronic self-monitored data (mood) on illness activity, would increase the accuracy as a marker of affective states

The pilot study included the monitoring of 28 outpatients with bipolar disorder, in a natural environment from October 2013 to December 2014. During this period, patients were given a smartphone application developed by Monsenso, to collect voice features, electronic self-monitoring data, and collection of automatically generated data.

  • Voice features
    The voice features were extracted from the patients’ phone calls throughout the day, using the open-source Media Interpretation by Large feature-space Extraction (openSMILE) toolkit, which is a feature extractor for signal processing and machine learning applications.
  • Electronic self-monitoring data
    Patients were requested to provide daily electronic self-monitoring data. The parameters evaluated included: mood, sleep length, medication intake, activity level, alcohol consumption, mixed mood, irritability, cognitive problems, stress levels, and individualised early warning signs.
  • Automatically generated data
    Through the smartphones’ sensors, automated data tracking different aspects of behavioural activities was collected on a daily basis. The data compiled by the smartphones included the number and duration of phone calls and text messages, accelerometer data, and phone usage.

Results

This innovative study revealed that changes in voice features could, in fact, detect individual changes in affective state. The accuracy of the prediction is increased, by combining voice features with automatically generated smartphone data on behavioural activities, and electronic self-monitoring. Therefore, according to the study, voice features collected by smartphones in a natural setting, could be used as an objective state marker in patients with bipolar disorder.

According to the researchers, the monitoring of symptoms in bipolar disorder and the accurate classification of affective states based exclusively on voice features has great potential.

Clinicians would be able to obtain accurate, objective data in real-time on the patients’ affective states based on collected voice features. The smartphone application could be used to monitor symptoms long-term, outside clinical settings, and enable early intervention between outpatient visits.

According to the researchers, at the time the study was conducted, it was the first study ever that investigated the combinations of voice features; automatically generated data, and electronic self-monitored data as state markers in patients with bipolar disorder. Using feature analysis collected in real-time from smartphones for classifying affective states in bipolar disorder reflects an innovative, objective and unobtrusive method for monitoring of illness activity (state) during long-term and in naturalistic settings.

Reference:
Voice analysis as an objective state marker in bipolar disorder. M. Faurholt-Jepsen, J.Busk, M.Frost, M.Vinberg, E.M.Christensen, O.Winther, J.E.Bardram and L.V.Kessing. 5 May 2016. http://www.nature.com/tp/journal/v6/n7/pdf/tp2016123a.pdf