Vocal acoustic features may be objective biomarkers of negative symptoms in schizophrenia: A cross-sectional study

Acoustic biomarkers for schizophrenia spectrum disorders and their associations with symptoms and cognitive functioning

BACKGROUNDS

Acoustic biomarkers for schizophrenia spectrum disorders (SSDs) hold great promise due to their capacity to capture emotional information, which is often impaired in these patients. These biomarkers are easily accessible, noninvasive, objective, and cost-effective. This study investigated the accuracy of different machine learning (ML) models in classifying patients with SSDs or schizophrenia (SZ) versus healthy controls (HCs), as well as patients with cognitive-deficit (Cog-D) versus cognitive-non-deficit (Cog-ND) versus HCs. Additionally, correlations of the top 25 features contributing to these classifications with psychopathology and cognitive functioning were explored.

METHODS

Speech data were collected from patients with SSDs (n = 238) and HCs (n = 157) using multiple tasks, including the reading of emotional sentences. The Extrapyramidal Symptom Rating Scale (ESRS) was used to control for potential medication effects on speech. Acoustic features were extracted using the openSMILE toolkit, and models were trained with 10-fold cross-validation. Partial correlation analysis, adjusted for ESRS and chlorpromazine (CPZ) equivalent, was conducted between the top 25 features and measures of psychopathology and cognitive functioning.

RESULTS

Among the five ML models, accuracy of support vector machine (SVM) model was the best. It classified SSDs versus HCs with 83 % accuracy when using all 7 tasks, and 85 % when using only the happy sentences task. The SVM classification accuracy for Cog-D versus Cog-ND within SSDs was poor across all tasks; however, the accuracy for Cog-D versus HCs was 79 % when using free speech or happy sentences. The accuracy for classifying SZ versus HCs and Cog-D versus Cog-ND versus HCs exhibited variations. Several of the top 25 acoustic features correlated significantly with attention and verbal memory in patients with SSDs.

CONCLUSIONS

Our findings suggested that acoustic analysis, combined with a ML approach, could be used to classify successfully SSDs or the Cog-D subtype versus HCs. Features related to pitch, loudness, and timbre were particularly associated with attention in patients with SSDs. Future research should explore further the potential applications of acoustic biomarkers in multi-class classification, treatment response, and relapse detection in patients with SSDs.

Exploring Voice Acoustic Features Associated with Cognitive Status in Korean Speakers: A Preliminary Machine Learning Study

Objective: To develop a non-invasive cognitive impairment detection system using speech data analysis, addressing the growing global dementia crisis and enabling accessible early screening through daily health monitoring. Methods: Speech data from 223 Korean patients were collected across eight tasks. Patients were classified based on Korean Mini-Mental State Examination scores. Four machine learning models were tested for three binary classification tasks. Voice acoustic features were extracted and analyzed. Results: The Deep Neural Network model performed best in two classification tasks, with Precision-Recall Area Under the Curve scores of 0.737 for severe vs. no impairment and 0.726 for mild vs. no impairment, while Random Forest achieved 0.715 for severe + mild vs. no impairment. Several acoustic features emerged as potentially important indicators, with DDA shimmer from the /i/ task and stdevF0 from the /puh-tuh-kuh/ task showing consistent patterns across classification tasks. Conclusions: This preliminary study suggests that certain acoustic features may be associated with cognitive status, though demographic factors significantly influence these relationships. Further research with demographically matched populations is needed to validate these findings.

Fundamental Frequency of the Voice in Schizophrenia and Its Value as a Biomarker of the Disease

Recent research on schizophrenia seeks to identify objective biomarkers of the disease. The voice, and in particular the fundamental frequency (F0), could be one of them.

METHODOLOGY

We conducted a cross-sectional and descriptive study with a sample of 154 people. Of these, 46 were diagnosed with schizophrenia, 41 were at substance abuse, and 67 formed the control group, matched in variables of sex, age, and educational level, but without substance use compared with the high-risk group.

RESULTS

The biomechanical analyses of the voice indicated significant differences between the groups, differentiated by gender: in men (F = 5.316; P = 0.006) and in women (F = 4.13; P = 0.004). The greatest differences between groups were observed in the group of vulnerable individuals, with some stability of the F0 in people with schizophrenia. Furthermore, we found correlations between positive symptoms and decreased F0 (r = -0.353; P = 0.016).

CONCLUSIONS

Our study shows that schizophrenia is associated with decreased F0 in both men and women, and that medication could stabilize this decrease. These findings have important implications for the objective monitoring and diagnosis of schizophrenia.

Task-voting for schizophrenia spectrum disorders prediction using machine learning across linguistic feature domains.. [DOI: 10.1101/2024.08.31.24312886] [Citation(s) in RCA: 0] [Impact Index Per Article: 0]

Background and Hypothesis

Identifying schizophrenia spectrum disorders (SSD) from spontaneous speech features is a key focus in computational psychiatry today.

Study Design

We present a task-voting procedure using different speech-elicitation tasks to predict SSD in Spanish, followed by ablation studies highlighting the roles of specific tasks and feature domains. Speech from five tasks was recorded from 92 subjects (49 with SSD and 41 controls). A total of 319 features were automatically extracted, from which 24 were pre-selected based on between-feature correlations and ANOVA F-values, covering acoustic-prosody, morphosyntax, and semantic similarity metrics.

Study Results

ExtraTrees-based classification using these features yielded an accuracy of 0.840 on hold-out data. Ablating picture descriptions impaired performance most, followed by story reading, retelling, and free speech. Removing morphosyntactic measures impaired performance most, followed by acoustic and semantic measures. Mixed-effect models suggested significant group differences on all 24 features. In SSD, speech patterns were slower and more variable temporally, while variations in pitch, amplitude, and sound intensity decreased. Semantic similarity between speech and prompts decreased, while minimal distances from embedding centroids to each word increased, and word-to-word similarity arrays became more predictable, all replicating patterns documented in other languages. Morphosyntactically, SSD patients used more first-person pronouns together with less third-person pronouns, and more punctuations and negations. Semantic metrics correlated with a range of positive symptoms, and multiple acoustic-prosodic features with negative symptoms.

Conclusions

This study highlights the importance of combining different speech tasks and features for SSD detection, and validates previously found patterns in psychosis for Spanish.

Evaluating the clinical utility of speech analysis and machine learning in schizophrenia: A pilot study

BACKGROUND

Schizophrenia is a serious mental disorder that significantly impacts social functioning and quality of life. However, current diagnostic methods lack objective biomarker support. While some studies have indicated differences in audio features between patients with schizophrenia and healthy controls, these findings are influenced by demographic information and variations in experimental paradigms. Therefore, it is crucial to explore stable and reliable audio biomarkers for an auxiliary diagnosis and disease severity prediction of schizophrenia.

METHOD

A total of 130 individuals (65 patients with schizophrenia and 65 healthy controls) read three fixed texts containing positive, neutral, and negative emotions, and recorded them. All audio signals were preprocessed and acoustic features were extracted by a librosa-0.9.2 toolkit. Independent sample t-tests were performed on two sets of acoustic features, and Pearson correlation on the acoustic features and Positive and Negative Syndrome Scale (PANSS) scores of the schizophrenia group. Classification algorithms in scikit-learn were used to diagnose schizophrenia and predict the level of negative symptoms.

RESULTS

Significant differences were observed between the two groups in the mfcc_8, mfcc_11, and mfcc_33 of mel-frequency cepstral coefficient (MFCC). Furthermore, a significant correlation was found between mfcc_7 and the negative PANSS scores. Through acoustic features, we could not only differentiate patients with schizophrenia from healthy controls with an accuracy of 0.815 but also predict the grade of the negative symptoms in schizophrenia with an average accuracy of 0.691.

CONCLUSIONS

The results demonstrated the considerable potential of acoustic characteristics as reliable biomarkers for diagnosing schizophrenia and predicting clinical symptoms.