Into the Wild: Transitioning from Recognizing Mood in Clinical Interactions to Personal Conversations for Individuals with Bipolar Disorder

Abstract

Bipolar Disorder, a mood disorder with recurrent mania and depression, requires ongoing monitoring and specialty management. Current monitoring strategies are clinically-based, engaging highly specialized medical professionals who are becoming increasingly scarce. Automatic speech-based monitoring via smartphones has the potential to augment clinical monitoring by providing inexpensive and unobtrusive measurements of a patient's daily life. The success of such an approach is contingent on the ability to successfully utilize "in-the-wild" data. However, most existing work on automatic mood detection uses datasets collected in clinical or laboratory settings. This study presents experiments in automatically detecting depression severity in individuals with Bipolar Disorder using data derived from clinical interviews and from personal conversations. We find that mood assessment is more accurate using data collected from clinical interactions, in part because of their highly structured nature. We demonstrate that although the features that are most effective in clinical interactions do not extend well to personal conversational data, we can identify alternative features relevant in personal conversational speech to detect mood symptom severity. Our results highlight the challenges unique to working with "in-the-wild" data, providing insight into the degree to which the predictive ability of speech features is preserved outside of a clinical interview.

Privacy Enhanced Multimodal Neural Representations for Emotion Recognition

“…The resulting predictions can then be used to bias response generation. Emotion classification is also used in mobile and web applications to identify heightened risk of suicidal ideation or mood fluctuations (Khorram et al ;Matton, McInnis, and Provost 2019;Gideon et al 2019), for the purpose of tracking or intervention. Data are sent from users' devices, including mobile applications (Khorram et al ) and Alexa or Google home devices (Piersol and Beddingfield 2019), and are stored on central servers for analysis.…”

How to Design a Relevant Corpus for Sleepiness Detection Through Voice?

“…On the other hand, spontaneous speech benefits from its proximity to ecological conditions and the analysis of the content of speech. Indeed, a recent study in depression detection has linked the augmentation of pronouns and negatively valenced language with the mood of patients affected by bipolar disorders, making it possible to detect through voice ( 70 ). Such a system could be developed for sleepiness and excessive sleepiness estimation with the detection of words such as “tired,” “exhausted,” and “sleepy.”…”

Modelling Paralinguistic Properties in Conversational Speech to Detect Bipolar Disorder and Borderline Personality Disorder

“…mean, max, variance, linear regression coefficients, etc.) for feature aggregation [9,10,11,12], where the role of these aggregation functions is to describe the global characteristics of given spoken conversation. However, the conversational dynamics is not effectively modelled during this process and important sequential information may be ignored as a result.…”