Automated Assessment of Psychiatric Disorders Using Speech: A Systematic Review

Abstract

Objective: There are many barriers to accessing mental health assessments including cost and stigma. Even when individuals receive professional care, assessments are intermittent and may be limited partly due to the episodic nature of psychiatric symptoms. Therefore, machine learning technology using speech samples obtained in the clinic or remotely could one day be a biomarker to improve diagnosis and treatment. To date, reviews have only focused on using acoustic features from speech to detect depression and schizophrenia. Here we present the first systematic review of studies using speech for automated assessments across a broader range of psychiatric disorders.Methods: We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. We included studies from the last 10 years using speech to identify the presence or severity of disorders within the Diagnostic and Statistical Manual of Mental Disorders (DSM–5). For each disorder we describe sample size, clinical evaluation method, speech-eliciting tasks, machine learning methodology, performance, and other relevant findings. Results: 1395 studies were screened of which 127 studies met the inclusion criteria. The majority of studies were on depression, schizophrenia, and bipolar disorder, and the remaining on post-traumatic stress disorder, anxiety disorders, and eating disorders. 63% of studies built machine learning predictive models, and the remaining 37% performed null-hypothesis testing only. We provide an online database with our search results and synthesize how acoustic features appear in each disorder.Conclusion: Speech processing technology could aid mental health assessments, but there are many obstacles to overcome, especially the need for comprehensive transdiagnostic and longitudinal studies. Given the diverse types of datasets, feature extraction, computational methodologies, and evaluation criteria, we provide guidelines for both acquiring data and building machine learning models with a focus on testing hypotheses, open science, reproducibility, and generalizability.

Voice Analysis for Neurological Disorder Recognition–A Systematic Review and Perspective on Emerging Trends

“…The aspect of which acoustic features are found to correlate with which neurological disorder was addressed prominently by Low et al for psychiatric disorders ( 17 ). In the broader scope of this review, we aimed to extend their synthesis to also incorporate acoustic features of the neurodegenerative disorders addressed in this review.…”

“…Further, there are features, which are tailored for disease-specific vocal dynamics [e.g., ( 8 ) on AD]. Low et al ( 17 ) provide a comprehensive overview of the commonly used acoustic features derived from speech in neurological disorder quantification. They regard the GeMAPS features and provide a glossary on the regarded features [based on Cummins et al ( 3 ) and Horwitz et al ( 18 )], to which we refer the interested reader.…”

“…Previous reviews in the field have summarized the state of voice analysis for individual disorders and a few reviews outlined the state of research across several neurological disorders. One prominent systematic review was performed by Low et al ( 17 ), in which they regarded a variety of psychiatric disorders (depression, PTSD, OCD, bulimia, anorexia, schizophrenia, hypomania, and anxiety). Therein, they synthesised which acoustic features are prominently changed in voice in each disorder.…”

“…However, the scope of the review of Low et al ( 17 ) was limited to psychiatric disorders and Voleti et al ( 28 ) did not perform a systematic literature search. In this context, a comprehensive review that provides a broad overview of the field of neurological disorder recognition from voice is needed.…”

“…In this context, a comprehensive review that provides a broad overview of the field of neurological disorder recognition from voice is needed. Therefore, we extended to the scope of Low et al ( 17 ) by also including the neurodegenerative disorders ALS, AD, MCI, MS, and PD. Further, we adopted a reproducible, systematic approach by querying bibliographic databases.…”