Prediction of psychosis across protocols and risk cohorts using automated language analysis

Assessing dimensions of thought disorder with large language models: The tradeoff of accuracy and consistency

Natural Language Processing (NLP) methods have shown promise for the assessment of formal thought disorder, a hallmark feature of schizophrenia in which disturbances to the structure, organization, or coherence of thought can manifest as disordered or incoherent speech. We investigated the suitability of modern Large Language Models (LLMs - e.g., GPT-3.5, GPT-4, and Llama 3) to predict expert-generated ratings for three dimensions of thought disorder (coherence, content, and tangentiality) assigned to speech samples collected from both patients with a diagnosis of schizophrenia (n = 26) and healthy control participants (n = 25). In addition to (1) evaluating the accuracy of LLM-generated ratings relative to human experts, we also (2) investigated the degree to which the LLMs produced consistent ratings across multiple trials, and we (3) sought to understand the factors that impacted the consistency of LLM-generated output. We found that machine-generated ratings of the level of thought disorder in speech matched favorably those of expert humans, and we identified a tradeoff between accuracy and consistency in LLM ratings. Unlike traditional NLP methods, LLMs were not always consistent in their predictions, but these inconsistencies could be mitigated with careful parameter selection and ensemble methods. We discuss implications for NLP-based assessment of thought disorder and provide recommendations of best practices for integrating these methods in the field of psychiatry.

Brain structural associations of syntactic complexity and diversity across schizophrenia spectrum and major depressive disorders, and healthy controls

Deviations in syntax production have been well documented in schizophrenia spectrum disorders (SSD). Recently, we have shown evidence for transdiagnostic subtypes of syntactic complexity and diversity. However, there is a lack of studies exploring brain structural correlates of syntax across diagnoses. We assessed syntactic complexity and diversity of oral language production using four Thematic Apperception Test pictures in a sample of N = 87 subjects (n = 24 major depressive disorder (MDD), n = 30 SSD patients both diagnosed according to DSM-IV-TR, and n = 33 healthy controls (HC)). General linear models were used to investigate the association of syntax with gray matter volume (GMV), fractional anisotropy (FA), axial (AD), radial (RD), and mean diffusivity (MD). Age, sex, total intracranial volume, group, interaction of group and syntax were covariates of no interest. Syntactic diversity was positively correlated with the GMV of the right medial pre- and postcentral gyri and with the FA of the left superior-longitudinal fasciculus (temporal part). Conversely, the AD of the left cingulum bundle and the forceps minor were negatively correlated with syntactic diversity. The AD of the right inferior-longitudinal fasciculus was positively correlated with syntactic complexity. Negative associations were observed between syntactic complexity and the FA of the left cingulum bundle, the right superior-longitudinal fasciculus, and the AD of the forceps minor and the left uncinate fasciculus. Our study showed brain structural correlates of syntactic complexity and diversity across diagnoses and HC. This contributes to a comprehensive understanding of the interplay between linguistic and neural substrates in syntax production in psychiatric disorders and HC.

Computational analysis of linguistic features in speech samples of first-episode bipolar disorder and psychosis

BACKGROUND

In recent years, automated analyses using novel NLP methods have been used to investigate language abnormalities in schizophrenia. In contrast, only a few studies used automated language analyses in bipolar disorder. To our knowledge, no previous research compared automated language characteristics of first-episode psychosis (FEP) and bipolar disorder (FEBD) using NLP methods.

METHODS

Our study included 53 FEP, 40 FEBD and 50 healthy control participants who are native Turkish speakers. Speech samples of the participants in the Thematic Apperception Test (TAT) underwent automated generic and part-of-speech analyses, as well as sentence-level semantic similarity analysis based on SBERT.

RESULTS

Both FEBD and FEP were associated with the use of shorter sentences and increased sentence-level semantic similarity but less semantic alignment with the TAT pictures. FEP also demonstrated reduced verbosity and syntactic complexity. FEP differed from FEBD in reduced verbosity, decreased first-person singular pronouns, fewer conjunctions, increased semantic similarity as well as shorter sentence and word length. The mean classification accuracy was 82.45 % in FEP vs HC, 71.1 % in FEBD vs HC, and 73 % in FEP vs FEBD. After Bonferroni correction, the severity of negative symptoms in FEP was associated with reduced verbal output and increased 5th percentile of semantic similarity.

LIMITATIONS

The main limitation of this study was the cross-sectional nature.

CONCLUSION

Our findings demonstrate that both patient groups showed language abnormalities, which were more severe and widespread in FEP compared to FEBD. Our results suggest that NLP methods reveal transdiagnostic linguistic abnormalities in FEP and FEBD.

Speech and language patterns in autism: Towards natural language processing as a research and clinical tool

Speech and language differences have long been described as important characteristics of autism spectrum disorder (ASD). Linguistic abnormalities range from prosodic differences in pitch, intensity, and rate of speech, to language idiosyncrasies and difficulties with pragmatics and reciprocal conversation. Heterogeneity of findings and a reliance on qualitative, subjective ratings, however, limit a full understanding of linguistic phenotypes in autism. This review summarizes evidence of both speech and language differences in ASD. We also describe recent advances in linguistic research, aided by automated methods and software like natural language processing (NLP) and speech analytic software. Such approaches allow for objective, quantitative measurement of speech and language patterns that may be more tractable and unbiased. Future research integrating both speech and language features and capturing "natural language" samples may yield a more comprehensive understanding of language differences in autism, offering potential implications for diagnosis, intervention, and research.

Automated linguistic analysis in youth at clinical high risk for psychosis

Identifying individuals at clinical high risk for psychosis (CHRP) is crucial for preventing psychosis and improving the prognosis for schizophrenia. Individuals at CHR-P may exhibit mild forms of formal thought disorder (FTD), making it possible to identify them using natural language processing (NLP) methods. In this study, speech samples of 62 CHR-P individuals and 45 healthy controls (HCs) were elicited using Thematic Apperception Test images. The evaluation involved various NLP measures such as semantic similarity, generic, and part-of-speech (POS) features. The CHR-P group demonstrated higher sentence-level semantic similarity and reduced mean image-to-text similarity. Regarding generic analysis, they demonstrated reduced verbosity and produced shorter sentences with shorter words. The POS analysis revealed a decrease in the utilization of adverbs, conjunctions, and first-person singular pronouns, alongside an increase in the utilization of adjectives in the CHR-P group compared to HC. In addition, we developed a machine-learning model based on 30 NLP-derived features to distinguish between the CHR-P and HC groups. The model demonstrated an accuracy of 79.6 % and an AUC-ROC of 0.86. Overall, these findings suggest that automated language analysis of speech could provide valuable information for characterizing FTD during the clinical high-risk phase and has the potential to be applied objectively for early intervention for psychosis.

Protocol for Designing a Model to Predict the Likelihood of Psychosis From Electronic Health Records Using Natural Language Processing and Machine Learning

INTRODUCTION

Rapid identification of individuals developing a psychotic spectrum disorder (PSD) is crucial because untreated psychosis is associated with poor outcomes and decreased treatment response. Lack of recognition of early psychotic symptoms often delays diagnosis, further worsening these outcomes.

METHODS

The proposed study is a cross-sectional, retrospective analysis of electronic health record data including clinician documentation and patient-clinician secure messages for patients aged 15-29 years with ≥ 1 primary care encounter between 2017 and 2019 within 2 Kaiser Permanente regions. Patients with new-onset PSD will be distinguished from those without a diagnosis if they have ≥ 1 PSD diagnosis within 12 months following the primary care encounter. The prediction model will be trained using a trisourced natural language processing feature extraction design and validated both within each region separately and in a modified combined sample.

DISCUSSION

This proposed model leverages the strengths of the large volume of patient-specific data from an integrated electronic health record with natural language processing to identify patients at elevated chance of developing a PSD. This project carries the potential to reduce the duration of untreated psychosis and thereby improve long-term patient outcomes.

Detecting schizophrenia, bipolar disorder, psychosis vulnerability and major depressive disorder from 5 minutes of online-collected speech

Background

Psychosis poses substantial social and healthcare burdens. The analysis of speech is a promising approach for the diagnosis and monitoring of psychosis, capturing symptoms like thought disorder and flattened affect. Recent advancements in Natural Language Processing (NLP) methodologies enable the automated extraction of informative speech features, which has been leveraged for early psychosis detection and assessment of symptomology. However, critical gaps persist, including the absence of standardized sample collection protocols, small sample sizes, and a lack of multi-illness classification, limiting clinical applicability. Our study aimed to (1) identify an optimal assessment approach for the online and remote collection of speech, in the context of assessing the psychosis spectrum and evaluate whether a fully automated, speech-based machine learning (ML) pipeline can discriminate among different conditions on the schizophrenia-bipolar spectrum (SSD-BD-SPE), help-seeking comparison subjects (MDD), and healthy controls (HC) at varying layers of analysis and diagnostic complexity.

Methods

We adopted online data collection methods to collect 20 minutes of speech and demographic information from individuals. Participants were categorized as "healthy" help-seekers (HC), having a schizophrenia-spectrum disorder (SSD), bipolar disorder (BD), major depressive disorder (MDD), or being on the psychosis spectrum with sub-clinical psychotic experiences (SPE). SPE status was determined based on self-reported clinical diagnosis and responses to the PHQ-8 and PQ-16 screening questionnaires, while other diagnoses were determined based on self-report from participants. Linguistic and paralinguistic features were extracted and ensemble learning algorithms (e.g., XGBoost) were used to train models. A 70%-30% train-test split and 30-fold cross-validation was used to validate the model performance.

Results

The final analysis sample included 1140 individuals and 22,650 minutes of speech. Using 5-minutes of speech, our model could discriminate between HC and those with a serious mental illness (SSD or BD) with 86% accuracy (AUC = 0.91, Recall = 0.7, Precision = 0.98). Furthermore, our model could discern among HC, SPE, BD and SSD groups with 86% accuracy (F1 macro = 0.855, Recall Macro = 0.86, Precision Macro = 0.86). Finally, in a 5-class discrimination task including individuals with MDD, our model had 76% accuracy (F1 macro = 0.757, Recall Macro = 0.758, Precision Macro = 0.766).

Conclusion

Our ML pipeline demonstrated disorder-specific learning, achieving excellent or good accuracy across several classification tasks. We demonstrated that the screening of mental disorders is possible via a fully automated, remote speech assessment pipeline. We tested our model on relatively high number conditions (5 classes) in the literature and in a stratified sample of psychosis spectrum, including HC, SPE, SSD and BD (4 classes). We tested our model on a large sample (N = 1150) and demonstrated best-in-class accuracy with remotely collected speech data in the psychosis spectrum, however, further clinical validation is needed to test the reliability of model performance.

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.

Deciphering language disturbances in schizophrenia: A study using fine-tuned language models

This research presents two stable language metrics, namely Successful Prediction Rate (SPR) and Disfluency (DF), to objectively quantify the linguistic disturbances associated with schizophrenia. These novel language metrics can capture both off-topic responses and incoherence in patients' speech by modeling speech information and fine-tuning techniques. Additionally, these metrics exhibit cultural sensitivity while providing a more comprehensive evaluation of linguistic abnormalities in schizophrenia. This research fine-tuned the ELECTRA Pretrained Language Model on a 750 MB text corpus obtained from major Chinese mental health forums. The effectiveness of the fine-tuned language model is verified on a group comprising 38 individuals diagnosed with schizophrenia and 25 meticulously matched healthy controls. The study explores the association between the fine-tuned language model and the Positive and Negative Syndrome Scale (PANSS) items. The results demonstrate that SPR is higher in healthy controls, indicating better language understanding by the pre-trained language model. Conversely, DF is higher in individuals with schizophrenia, indicating more inconsistent language structure. The relationship between linguistic features and P2 (conceptual disorganization) reveals that patients with positive P2 exhibit lower SPR and higher DF. Binary logistic regression using the combined SPR and DF features achieves 84.5 % accuracy in classifying P2, exceeding the performance of traditional features by 20.5 %. Moreover, the proposed linguistic features outperform traditional linguistic features in discriminating FTD (formal thought disorder), as demonstrated by multivariate linear regression analysis.

Language in interaction: turn-taking patterns in conversations involving individuals with schizophrenia

Individuals with schizophrenia generally show difficulties in interpersonal communication. Linguistic analyses shed new light on speech atypicalities in schizophrenia. However, very little is known about conversational interaction management by these individuals. Moreover, the relationship between linguistic features, psychopathology, and patients' subjectivity has received limited attention to date. We used a novel methodology to explore dyadic conversations involving 58 participants (29 individuals with schizophrenia and 29 control persons) and medical doctors. High-quality stereo recordings were obtained and used to quantify turn-taking patterns. We investigated psychopathological dimensions and subjective experiences using the Positive and Negative Syndrome Scale for Schizophrenia (PANSS), the Examination of Anomalous Self Experience scale (EASE), the Autism Rating Scale (ARS) and the Abnormal Bodily Phenomena questionnaire (ABPq). Different turn-taking patterns of both patients and interviewers characterised conversations involving individuals with schizophrenia. We observed higher levels of overlap and mutual silence in dialogues with the patients compared to dialogues with control persons. Mutual silence was associated with negative symptom severity; no dialogical feature was correlated with anomalous subjective experiences. Our findings suggest that individuals with schizophrenia display peculiar turn-taking behaviour, thereby enhancing our understanding of interactional coordination in schizophrenia.

Automated Speech Analysis in Bipolar Disorder: The CALIBER Study Protocol and Preliminary Results

Background: Bipolar disorder (BD) involves significant mood and energy shifts reflected in speech patterns. Detecting these patterns is crucial for diagnosis and monitoring, currently assessed subjectively. Advances in natural language processing offer opportunities to objectively analyze them. Aims: To (i) correlate speech features with manic-depressive symptom severity in BD, (ii) develop predictive models for diagnostic and treatment outcomes, and (iii) determine the most relevant speech features and tasks for these analyses. Methods: This naturalistic, observational study involved longitudinal audio recordings of BD patients at euthymia, during acute manic/depressive phases, and after-response. Patients participated in clinical evaluations, cognitive tasks, standard text readings, and storytelling. After automatic diarization and transcription, speech features, including acoustics, content, formal aspects, and emotionality, will be extracted. Statistical analyses will (i) correlate speech features with clinical scales, (ii) use lasso logistic regression to develop predictive models, and (iii) identify relevant speech features. Results: Audio recordings from 76 patients (24 manic, 21 depressed, 31 euthymic) were collected. The mean age was 46.0 ± 14.4 years, with 63.2% female. The mean YMRS score for manic patients was 22.9 ± 7.1, reducing to 5.3 ± 5.3 post-response. Depressed patients had a mean HDRS-17 score of 17.1 ± 4.4, decreasing to 3.3 ± 2.8 post-response. Euthymic patients had mean YMRS and HDRS-17 scores of 0.97 ± 1.4 and 3.9 ± 2.9, respectively. Following data pre-processing, including noise reduction and feature extraction, comprehensive statistical analyses will be conducted to explore correlations and develop predictive models. Conclusions: Automated speech analysis in BD could provide objective markers for psychopathological alterations, improving diagnosis, monitoring, and response prediction. This technology could identify subtle alterations, signaling early signs of relapse. Establishing standardized protocols is crucial for creating a global speech cohort, fostering collaboration, and advancing BD understanding.

Predicting treatment resistance in schizophrenia patients: Machine learning highlights the role of early pathophysiologic features

Detecting patients with a high-risk profile for treatment-resistant schizophrenia (TRS) can be beneficial for implementing individually adapted therapeutic strategies and better understanding the TRS etiology. The aim of this study was to explore, with machine learning methods, the impact of demographic and clinical patient characteristics on TRS prediction, for already established risk factors and unexplored ones. This was a retrospective study of 500 patients admitted during 2020 to the University Hospital Group for Paris Psychiatry. We hypothesized potential TRS risk factors. The selected features were coded into structured variables in a new dataset, by processing patients discharge summaries and medical narratives with natural-language processing methods. We compared three machine learning models (XGBoost, logistic elastic net regression, logistic regression without regularization) for predicting TRS outcome. We analysed feature impact on the models, suggesting the following factors as markers of a high-risk TRS profile: early age at first contact with psychiatry, antipsychotic treatment interruptions due to non-adherence, absence of positive symptoms at baseline, educational problems and adolescence mental disorders in the personal psychiatric history. Specifically, we found a significant association with TRS outcome for age at first contact with psychiatry and medication non-adherence. Our findings on TRS risk factors are consistent with the review of the literature and suggest potential in using early pathophysiologic features for TRS prediction. Results were encouraging with the use of natural-langage processing techniques to leverage raw data provided by discharge summaries, combined with machine leaning models. These findings are a promising step for helping clinicians adapt their guidelines to early detection of TRS.

Natural Language Processing and Schizophrenia: A Scoping Review of Uses and Challenges

(1) Background: Approximately 1% of the global population is affected by schizophrenia, a disorder marked by cognitive deficits, delusions, hallucinations, and language issues. It is associated with genetic, neurological, and environmental factors, and linked to dopaminergic hyperactivity and neurotransmitter imbalances. Recent research reveals that patients exhibit significant language impairments, such as reduced verbal output and fluency. Advances in machine learning and natural language processing show potential for early diagnosis and personalized treatments, but additional research is required for the practical application and interpretation of such technology. The objective of this study is to explore the applications of natural language processing in patients diagnosed with schizophrenia. (2) Methods: A scoping review was conducted across multiple electronic databases, including Medline, PubMed, Embase, and PsycInfo. The search strategy utilized a combination of text words and subject headings, focusing on schizophrenia and natural language processing. Systematically extracted information included authors, population, primary uses of the natural language processing algorithms, main outcomes, and limitations. The quality of the identified studies was assessed. (3) Results: A total of 516 eligible articles were identified, from which 478 studies were excluded based on the first analysis of titles and abstracts. Of the remaining 38 studies, 18 were selected as part of this scoping review. The following six main uses of natural language processing were identified: diagnostic and predictive modeling, followed by specific linguistic phenomena, speech and communication analysis, social media and online content analysis, clinical and cognitive assessment, and linguistic feature analysis. (4) Conclusions: This review highlights the main uses of natural language processing in the field of schizophrenia and the need for more studies to validate the effectiveness of natural language processing in diagnosing and treating schizophrenia.

Clinical efficacy of pre-trained large language models through the lens of aphasia

The rapid development of large language models (LLMs) motivates us to explore how such state-of-the-art natural language processing systems can inform aphasia research. What kind of language indices can we derive from a pre-trained LLM? How do they differ from or relate to the existing language features in aphasia? To what extent can LLMs serve as an interpretable and effective diagnostic and measurement tool in a clinical context? To investigate these questions, we constructed predictive and correlational models, which utilize mean surprisals from LLMs as predictor variables. Using AphasiaBank archived data, we validated our models' efficacy in aphasia diagnosis, measurement, and prediction. Our finding is that LLMs-surprisals can effectively detect the presence of aphasia and different natures of the disorder, LLMs in conjunction with the existing language indices improve models' efficacy in subtyping aphasia, and LLMs-surprisals can capture common agrammatic deficits at both word and sentence level. Overall, LLMs have potential to advance automatic and precise aphasia prediction. A natural language processing pipeline can be greatly benefitted from integrating LLMs, enabling us to refine models of existing language disorders, such as aphasia.

Reduced speech coherence in psychosis-related social media forum posts

The extraction of linguistic markers from social media posts, which are indicative of the onset and course of mental disorders, offers great potential for mental healthcare. In the present study, we extracted over one million posts from the popular social media platform Reddit to analyze speech coherence, which reflects formal thought disorder and is a characteristic feature of schizophrenia and associated psychotic disorders. Natural language processing (NLP) models were used to perform an automated quantification of speech coherence. We could demonstrate that users who are active on forums geared towards disorders with a higher degree of psychotic symptoms tend to show a lower level of coherence. The lowest coherence scores were found in users of forums on dissociative identity disorder, schizophrenia, and bipolar disorder. In contrast, a relatively high level of coherence was detected in users of forums related to obsessive-compulsive disorder, anxiety, and depression. Users of forums on posttraumatic stress disorder, autism, and attention-deficit hyperactivity disorder exhibited medium-level coherence. Our findings provide promising first evidence for the possible utility of NLP-based coherence analyses for the early detection and prevention of psychosis on the basis of posts gathered from publicly available social media data. This opens new avenues for large-scale prevention programs aimed at high-risk populations.

Interdisciplinary approach to identify language markers for post-traumatic stress disorder using machine learning and deep learning

Post-traumatic stress disorder (PTSD) lacks clear biomarkers in clinical practice. Language as a potential diagnostic biomarker for PTSD is investigated in this study. We analyze an original cohort of 148 individuals exposed to the November 13, 2015, terrorist attacks in Paris. The interviews, conducted 5-11 months after the event, include individuals from similar socioeconomic backgrounds exposed to the same incident, responding to identical questions and using uniform PTSD measures. Using this dataset to collect nuanced insights that might be clinically relevant, we propose a three-step interdisciplinary methodology that integrates expertise from psychiatry, linguistics, and the Natural Language Processing (NLP) community to examine the relationship between language and PTSD. The first step assesses a clinical psychiatrist's ability to diagnose PTSD using interview transcription alone. The second step uses statistical analysis and machine learning models to create language features based on psycholinguistic hypotheses and evaluate their predictive strength. The third step is the application of a hypothesis-free deep learning approach to the classification of PTSD in our cohort. Results show that the clinical psychiatrist achieved a diagnosis of PTSD with an AUC of 0.72. This is comparable to a gold standard questionnaire (Area Under Curve (AUC) ≈ 0.80). The machine learning model achieved a diagnostic AUC of 0.69. The deep learning approach achieved an AUC of 0.64. An examination of model error informs our discussion. Importantly, the study controls for confounding factors, establishes associations between language and DSM-5 subsymptoms, and integrates automated methods with qualitative analysis. This study provides a direct and methodologically robust description of the relationship between PTSD and language. Our work lays the groundwork for advancing early and accurate diagnosis and using linguistic markers to assess the effectiveness of pharmacological treatments and psychotherapies.

Emotional tone in clinical high risk for psychosis: novel insights from a natural language analysis approach

Background

Individuals at clinical high risk (CHR) for psychosis experience subtle emotional disturbances that are traditionally difficult to assess, but natural language processing (NLP) methods may provide novel insight into these symptoms. We predicted that CHR individuals would express more negative emotionality and less emotional language when compared to controls. We also examined associations with symptomatology.

Methods

Participants included 49 CHR individuals and 42 healthy controls who completed a semi-structured narrative interview. Interview transcripts were analyzed using Linguistic Inquiry and Word Count (LIWC) to assess the emotional tone of the language (tone -the ratio of negative to positive language) and count positive/negative words used. Participants also completed clinical symptom assessments to determine CHR status and characterize symptoms (i.e., positive and negative symptom domains).

Results

The CHR group had more negative emotional tone compared to healthy controls (t=2.676, p=.009), which related to more severe positive symptoms (r2=.323, p=.013). The percentages of positive and negative words did not differ between groups (p's>.05).

Conclusions

Language analyses provided accessible, ecologically valid insight into affective dysfunction and psychosis risk symptoms. Natural language processing analyses unmasked differences in language for CHR that captured language tendencies that were more nuanced than the words that are chosen.

Automated linguistic analysis in speech samples of Turkish-speaking patients with schizophrenia-spectrum disorders

Modern natural language processing (NLP) methods provide ways to objectively quantify language disturbances for potential use in diagnostic classification. We performed computerized language analysis in speech samples of 82 Turkish-speaking subjects, including 44 patients with schizophrenia spectrum disorders (SSD) and 38 healthy controls (HC). Exploratory analysis of speech samples involved 16 sentence-level semantic similarity features using SBERT (Sentence Bidirectional Encoder Representation from Text) as well as 8 generic and 8 part-of-speech (POS) features. The random forest classifier using SBERT-derived semantic similarity features achieved a mean accuracy of 85.6 % for the classification of SSD and HC. When semantic similarity features were combined with generic and POS features, the classifier's mean accuracy reached to 86.8 %. Our analysis reflected increased sentence-level semantic similarity scores in SSD. Generic and POS analyses revealed an increase in the use of verbs, proper nouns and pronouns in SSD while our results showed a decrease in the utilization of conjunctions, determiners, and both average and maximum sentence length in SSD compared to HC. Quantitative language features were correlated with the expressive deficit domain of BNSS (Brief Negative Symptom Scale) as well as with the duration of illness. These findings from Turkish-speaking interviews contribute to the growing evidence-based NLP-derived assessments in non-English-speaking patients.

Accelerating Medicines Partnership® Schizophrenia (AMP® SCZ): Rationale and Study Design of the Largest Global Prospective Cohort Study of Clinical High Risk for Psychosis

This article describes the rationale, aims, and methodology of the Accelerating Medicines Partnership® Schizophrenia (AMP® SCZ). This is the largest international collaboration to date that will develop algorithms to predict trajectories and outcomes of individuals at clinical high risk (CHR) for psychosis and to advance the development and use of novel pharmacological interventions for CHR individuals. We present a description of the participating research networks and the data processing analysis and coordination center, their processes for data harmonization across 43 sites from 13 participating countries (recruitment across North America, Australia, Europe, Asia, and South America), data flow and quality assessment processes, data analyses, and the transfer of data to the National Institute of Mental Health (NIMH) Data Archive (NDA) for use by the research community. In an expected sample of approximately 2000 CHR individuals and 640 matched healthy controls, AMP SCZ will collect clinical, environmental, and cognitive data along with multimodal biomarkers, including neuroimaging, electrophysiology, fluid biospecimens, speech and facial expression samples, novel measures derived from digital health technologies including smartphone-based daily surveys, and passive sensing as well as actigraphy. The study will investigate a range of clinical outcomes over a 2-year period, including transition to psychosis, remission or persistence of CHR status, attenuated positive symptoms, persistent negative symptoms, mood and anxiety symptoms, and psychosocial functioning. The global reach of AMP SCZ and its harmonized innovative methods promise to catalyze the development of new treatments to address critical unmet clinical and public health needs in CHR individuals.

A new schizophrenia screening instrument based on evaluating the patient's writing

Formal Thought Disorder (FTD) is a defining feature of schizophrenia, which is often assessed through patients' speech. Meanwhile, the written language is less studied. The aim of the present study is to establish and validate a comprehensive clinical screening scale, capturing the full variety of empirical characteristics of writing in patients with schizophrenia. The 16-item Screening Instrument for Schizophrenic Features in Writing (SISFiW) is derived from detailed literature review and a "brainstorming" discussion on 30 samples written by patients with schizophrenia. One hundred and fifty-seven participants (114 patients with an ICD-10 diagnoses of schizophrenia; 43 healthy control subjects) were interviewed and symptoms assessed with the Positive and Negative Syndrome Scale (PANSS) and the Scale for the Assessment of Thought, Language, and Communication (TLC). Article samples written by each participant were rated with the SISFiW. Results demonstrated significant difference of the SISFiW-total between the patient group and healthy controls [(3.61 ± 1.72) vs. (0.49 ± 0.63), t = 16.64, p<0.001]. The inter-rater reliability (weighted kappa = 0.72) and the internal consistency (Cronbach's alpha coefficient = 0.613) were acceptable, but correlations with the criterion (PANSS and TLC) were unremarkable. The ROC analysis indicated a cutoff point at 2 with the maximal sensitivity (93.0 %)/specificity (93.0 %). Discriminant analysis of the SISFiW items yielded 8 classifiers that discriminated between the diagnostic groups at a perfect overall performance (with 90.4 % of original and 88.5 % cross-validated grouped cases classified correctly). This instrument appears to be practicable and reliable, with relatively robust discriminatory power, and may serve as a complementary tool to existing FTD rating scales.

Speech markers to predict and prevent recurrent episodes of psychosis: A narrative overview and emerging opportunities

Preventing relapse in schizophrenia improves long-term health outcomes. Repeated episodes of psychotic symptoms shape the trajectory of this illness and can be a detriment to functional recovery. Despite early intervention programs, high relapse rates persist, calling for alternative approaches in relapse prevention. Predicting imminent relapse at an individual level is critical for effective intervention. While clinical profiles are often used to foresee relapse, they lack the specificity and sensitivity needed for timely prediction. Here, we review the use of speech through Natural Language Processing (NLP) to predict a recurrent psychotic episode. Recent advancements in NLP of speech have shown the ability to detect linguistic markers related to thought disorder and other language disruptions within 2-4 weeks preceding a relapse. This approach has shown to be able to capture individual speech patterns, showing promise in its use as a prediction tool. We outline current developments in remote monitoring for psychotic relapses, discuss the challenges and limitations and present the speech-NLP based approach as an alternative to detect relapses with sufficient accuracy, construct validity and lead time to generate clinical actions towards prevention.

Combining Clinical With Cognitive or Magnetic Resonance Imaging Data for Predicting Transition to Psychosis in Ultra High-Risk Patients: Data From the PACE 400 Cohort

BACKGROUND

Multimodal modeling that combines biological and clinical data shows promise in predicting transition to psychosis in individuals who are at ultra-high risk. Individuals who transition to psychosis are known to have deficits at baseline in cognitive function and reductions in gray matter volume in multiple brain regions identified by magnetic resonance imaging.

METHODS

In this study, we used Cox proportional hazards regression models to assess the additive predictive value of each modality-cognition, cortical structure information, and the neuroanatomical measure of brain age gap-to a previously developed clinical model using functioning and duration of symptoms prior to service entry as predictors in the Personal Assessment and Crisis Evaluation (PACE) 400 cohort. The PACE 400 study is a well-characterized cohort of Australian youths who were identified as ultra-high risk of transitioning to psychosis using the Comprehensive Assessment of At Risk Mental States (CAARMS) and followed for up to 18 years; it contains clinical data (from N = 416 participants), cognitive data (n = 213), and magnetic resonance imaging cortical parameters extracted using FreeSurfer (n = 231).

RESULTS

The results showed that neuroimaging, brain age gap, and cognition added marginal predictive information to the previously developed clinical model (fraction of new information: neuroimaging 0%-12%, brain age gap 7%, cognition 0%-16%).

CONCLUSIONS

In summary, adding a second modality to a clinical risk model predicting the onset of a psychotic disorder in the PACE 400 cohort showed little improvement in the fit of the model for long-term prediction of transition to psychosis.

Natural language processing for defining linguistic features in schizophrenia: A sample from Turkish speakers

Natural language processing (NLP) provides fast and accurate extraction of features related to the language of schizophrenia. We utilized NLP methods to test the hypothesis that schizophrenia is associated with altered linguistic features in Turkish, a non-Indo-European language, compared to controls. We also explored whether these possible altered linguistic features were language-dependent or -independent. We extracted and compared speech in schizophrenia (SZ, N = 38) and healthy well-matched control (HC, N = 38) participants using NLP. The analysis was conducted in two parts. In the first one, mean sentence length, total completed words, moving average type-token ratio to measure the lexical diversity, and first-person singular pronoun usage were calculated. In the second one, we used parts-of-speech tagging (POS) and Word2Vec in schizophrenia and control. We found that SZ had lower mean sentence length and moving average type-token ratio but higher use of first-person singular pronoun. All these significant results were correlated with the Thought and Language Disorder Scale score. The POS approach demonstrated that SZ used fewer coordinating conjunctions. Our methodology using Word2Vec detected that SZ had higher semantic similarity than HC and K-Means could differentiate between SZ and HC into two distinct groups with high accuracy, 86.84 %. Our findings showed that altered linguistic features in SZ are mostly language-independent. They are promising to describe language patterns in schizophrenia which proposes that NLP measurements may allow for rapid and objective measurements of linguistic features.

Towards a scalable approach to assess speech organization across the psychosis-spectrum -online assessment in conjunction with automated transcription and extraction of speech measures

Automatically extracted measures of speech constitute a promising marker of psychosis as disorganized speech is associated with psychotic symptoms and predictive of psychosis-onset. The potential of speech markers is, however, hampered by (i) lengthy assessments in laboratory settings and (ii) manual transcriptions. We investigated whether a short, scalable data collection (online) and processing (automated transcription) procedure would provide data of sufficient quality to extract previously validated speech measures. To evaluate the fit of our approach for purpose, we assessed speech in relation to psychotic-like experiences in the general population. Participants completed an 8-minute-long speech task online. Sample 1 included measures of psychometric schizotypy and delusional ideation (N = 446). Sample 2 included a low and high psychometric schizotypy group (N = 144). Recordings were transcribed both automatically and manually, and connectivity, semantic, and syntactic speech measures were extracted for both types of transcripts. 73%/86% participants in sample 1/2 completed the experiment. Nineteen out of 25 speech measures were strongly (r > 0.7) and significantly correlated between automated and manual transcripts in both samples. Amongst the 14 connectivity measures, 11 showed a significant relationship with delusional ideation. For the semantic and syntactic measures, On Topic score and the Frequency of personal pronouns were negatively correlated with both schizotypy and delusional ideation. Combined with demographic information, the speech markers could explain 11-14% of the variation of delusional ideation and schizotypy in Sample 1 and could discriminate between high-low schizotypy with high accuracy (0.72-0.70, AUC = 0.78-0.79) in Sample 2. The moderate to high retention rate, strong correlation of speech measures across manual and automated transcripts and sensitivity to psychotic-like experiences provides initial evidence that online collected speech in combination with automatic transcription is a feasible approach to increase accessibility and scalability of speech-based assessment of psychosis.

Navigating the semantic space: Unraveling the structure of meaning in psychosis using different computational language models

Speech in psychosis has long been ascribed as involving 'loosening of associations'. We pursued the aim to elucidate its underlying cognitive mechanisms by analysing picture descriptions from 94 subjects (29 healthy controls, 18 participants at clinical high risk, 29 with first-episode psychosis, and 18 with chronic schizophrenia), using five language models with different computational architectures: FastText, which represents meaning non-contextually/statically; BERT, which represents contextual meaning sensitive to grammar and context; Infersent and SBERT, which provide sentential representations; and CLIP, which evaluates speech relative to a visual stimulus. These models were used to quantify semantic distances crossed between successive tokens/sentences, and semantic perplexity indicating unexpectedness in continuations. Results showed that, among patients, semantic similarity increased when measured with FastText, Infersent, and SBERT, while it decreased with CLIP and BERT. Higher perplexity was observed in first-episode psychosis. Static semantic measures were associated with clinically measured impoverishment of thought and referential semantic measures with disorganization. These patterns indicate a shrinking conceptual semantic space as represented by static language models, which co-occurs with a widening in the referential semantic space as represented by contextual models. This duality underlines the need to separate these two forms of meaning for understanding mechanisms involved in semantic change in psychosis.

Meta-analysis of the relationship between metacognition and disorganized symptoms in psychosis

OBJECTIVE

Disorganized symptoms show associations with metacognitive deficits in psychosis. However, the magnitude of this relationship is unclear. This meta-analysis aimed to 1) quantify relationships between metacognition and both disorganized symptoms and disorganized speech; and 2) examine moderators of these relationships (e.g., metacognition type, neurocognition).

METHOD

A literature search was conducted using PsycINFO, Web of Science, PubMed, and EMBASE databases. English-language studies measuring disorganized symptoms and metacognition (i.e., introspective accuracy, metacognitive beliefs, or metacognitive capacity) in psychosis were included. Random effects meta-analyses were conducted using Pearson's r.

RESULTS

Meta-analysis of 20 studies (n = 1490) resulted in a significant negative medium correlation between disorganized symptoms and metacognition (r = -0.332, 95 % CI [-0.423, -0.235]). Magnitude was moderated by metacognition type. A significant negative small correlation between disorganized speech and metacognition (r = -0.173, 95 % CI [-0.254, -0.089], n = 1470) was observed, with no significant moderators.

CONCLUSIONS

Results clarify the magnitude of the relationships between metacognition and both disorganized symptoms and disorganized speech. Significant relationships may indicate conceptual links, yet the different magnitudes may reflect a distinction between disorganized symptoms and speech. The moderator finding highlights that metacognitive capacity has an especially strong link to disorganized symptoms and underscores the need for careful distinction between types of metacognition in future work.

Automated analysis of speech as a marker of sub-clinical psychotic experiences

Automated speech analysis techniques, when combined with artificial intelligence and machine learning, show potential in capturing and predicting a wide range of psychosis symptoms, garnering attention from researchers. These techniques hold promise in predicting the transition to clinical psychosis from at-risk states, as well as relapse or treatment response in individuals with clinical-level psychosis. However, challenges in scientific validation hinder the translation of these techniques into practical applications. Although sub-clinical research could aid to tackle most of these challenges, there have been only few studies conducted in speech and psychosis research in non-clinical populations. This work aims to facilitate this work by summarizing automated speech analytical concepts and the intersection of this field with psychosis research. We review psychosis continuum and sub-clinical psychotic experiences, and the benefits of researching them. Then, we discuss the connection between speech and psychotic symptoms. Thirdly, we overview current and state-of-the art approaches to the automated analysis of speech both in terms of language use (text-based analysis) and vocal features (audio-based analysis). Then, we review techniques applied in subclinical population and findings in these samples. Finally, we discuss research challenges in the field, recommend future research endeavors and outline how research in subclinical populations can tackle the listed challenges.

Language and communication rehabilitation in patients with schizophrenia: A narrative review

Language impairments often appear in patients with schizophrenia and are potential targets for rehabilitation. Clinical practice and research should be intimately connected. The aim was to perform a narrative review of the assessment and intervention tools that have been used for the rehabilitation of schizophrenia patients with language and communication impairments. Two types of tools, general and specific, were developed for both purposes. General tools include the Positive and Negative Syndrome Scale for assessment, and the Integrated Psychological Therapy for intervention. The specific tools used to evaluate language and communication impairments include the Scale for the Assessment of Thought, Language and Communication, the Formal Thought Disorder scales (for caregivers and patients), and the Thought and Language Disorder scale. The most recent language-specific intervention tools include the Cognitive Pragmatic Treatment, Conecta-2, Let's talk! Multimodal Speech-Gesture training, Speech Therapy Intervention Group, and PragmaCom. These tools primarily involve psychopathology/psychiatry, psychology, linguistics, speech and language therapy, and nursing. In conclusion, a wide range of assessment and intervention tools are available for the rehabilitation of language and communication impairments associated with schizophrenia. An integrative and interdisciplinary approach should always be considered for rehabilitation of language and communication in patients with schizophrenia throughout their lifetime.

Eye Movement Characteristics for Predicting a Transition to Psychosis: Longitudinal Changes and Implications

BACKGROUND AND HYPOTHESIS

Substantive inquiry into the predictive power of eye movement (EM) features for clinical high-risk (CHR) conversion and their longitudinal trajectories is currently sparse. This study aimed to investigate the efficiency of machine learning predictive models relying on EM indices and examine the longitudinal alterations of these indices across the temporal continuum.

STUDY DESIGN

EM assessments (fixation stability, free-viewing, and smooth pursuit tasks) were performed on 140 CHR and 98 healthy control participants at baseline, followed by a 1-year longitudinal observational study. We adopted Cox regression analysis and constructed random forest prediction models. We also employed linear mixed-effects models (LMMs) to analyze longitudinal changes of indices while stratifying by group and time.

STUDY RESULTS

Of the 123 CHR participants who underwent a 1-year clinical follow-up, 25 progressed to full-blown psychosis, while 98 remained non-converters. Compared with the non-converters, the converters exhibited prolonged fixation durations, decreased saccade amplitudes during the free-viewing task; larger saccades, and reduced velocity gain during the smooth pursuit task. Furthermore, based on 4 baseline EM measures, a random forest model classified converters and non-converters with an accuracy of 0.776 (95% CI: 0.633, 0.882). Finally, LMMs demonstrated no significant longitudinal alterations in the aforementioned indices among converters after 1 year.

CONCLUSIONS

Aberrant EMs may precede psychosis onset and remain stable after 1 year, and applying eye-tracking technology combined with a modeling approach could potentially aid in predicting CHRs evolution into overt psychosis.

Identification of risk factors for the onset of delirium associated with COVID-19 by mining nursing records

COVID-19 has a range of complications, from no symptoms to severe pneumonia. It can also affect multiple organs including the nervous system. COVID-19 affects the brain, leading to neurological symptoms such as delirium. Delirium, a sudden change in consciousness, can increase the risk of death and prolong the hospital stay. However, research on delirium prediction in patients with COVID-19 is insufficient. This study aimed to identify new risk factors that could predict the onset of delirium in patients with COVID-19 using machine learning (ML) applied to nursing records. This retrospective cohort study used natural language processing and ML to develop a model for classifying the nursing records of patients with delirium. We extracted the features of each word from the model and grouped similar words. To evaluate the usefulness of word groups in predicting the occurrence of delirium in patients with COVID-19, we analyzed the temporal changes in the frequency of occurrence of these word groups before and after the onset of delirium. Moreover, the sensitivity, specificity, and odds ratios were calculated. We identified (1) elimination-related behaviors and conditions and (2) abnormal patient behavior and conditions as risk factors for delirium. Group 1 had the highest sensitivity (0.603), whereas group 2 had the highest specificity and odds ratio (0.938 and 6.903, respectively). These results suggest that these parameters may be useful in predicting delirium in these patients. The risk factors for COVID-19-associated delirium identified in this study were more specific but less sensitive than the ICDSC (Intensive Care Delirium Screening Checklist) and CAM-ICU (Confusion Assessment Method for the Intensive Care Unit). However, they are superior to the ICDSC and CAM-ICU because they can predict delirium without medical staff and at no cost.

Prosody and schizophrenia. Objective acoustic measurements of monotonous and flat intonation in young Danish people with a schizophrenia diagnosis. A pilot study

PURPOSE

Patients with schizophrenia have a flat and monotonous intonation. The purpose of the study was to find the variables of flat speech that differed in patients from those in healthy controls in Danish.

MATERIALS AND METHODS

We compared drug-naïve schizophrenic patients 5 men, 13 women and 18 controls, aged 18-35 years, which had all grown up in Copenhagen speaking modern Danish standard (rigsdansk). We used two different tasks that lay different demands on the speaker to elicit spontaneous speech: a retelling of a film clip and telling a story from pictures in a book. A linguist used the computer program Praat to extract the phonetic linguistic parameters.

RESULTS

We found different results for the two elicitation tasks (Task 1: a retelling of a film clip, task 2: telling a story from pictures in a book). There was higher intensity variation in task one in controls and higher pitch variation in task two in controls. We found a difference in intensity with higher intensity variation in the stresses in the controls in task one and fewer syllables between each stress in the controls. We also found higher F1 variation in task one and two in the patient group and higher F2 variation in the control group in both tasks.

CONCLUSIONS

The results varied between patients and controls, but the demands also made a difference. Further research is needed to elucidate the possibilities of acoustic measures in diagnostics or linguistic treatment related to schizophrenia.

How Can Digital Mental Health Enhance Psychiatry?

The use of digital technologies is constantly growing around the world. The wider-spread adoption of digital technologies and solutions in the daily clinical practice in psychiatry seems to be a question of when, not if. We propose a synthesis of the scientific literature on digital technologies in psychiatry and discuss the main aspects of its possible uses and interests in psychiatry according to three domains of influence that appeared to us: 1) assist and improve current care: digital psychiatry allows for more people to have access to care by simply being more accessible but also by being less stigmatized and more convenient; 2) develop new treatments: digital psychiatry allows for new treatments to be distributed via apps, and practical guidelines can reduce ethical challenges and increase the efficacy of digital tools; and 3) produce scientific and medical knowledge: digital technologies offer larger and more objective data collection, allowing for more detection and prevention of symptoms. Finally, ethical and efficacy issues remain, and some guidelines have been put forth on how to safely use these solutions and prepare for the future.

Linguistic findings in persons with schizophrenia-a review of the current literature

INTRODUCTION

Alterations of verbalized thought occur frequently in psychotic disorders. We characterize linguistic findings in individuals with schizophrenia based on the current literature, including findings relevant for differential and early diagnosis.

METHODS

Review of literature published via PubMed search between January 2010 and May 2022.

RESULTS

A total of 143 articles were included. In persons with schizophrenia, language-related alterations can occur at all linguistic levels. Differentiating from findings in persons with affective disorders, typical symptoms in those with schizophrenia mainly include so-called "poverty of speech," reduced word and sentence production, impaired processing of complex syntax, pragmatic language deficits as well as reduced semantic verbal fluency. At the at-risk state, "poverty of content," pragmatic difficulties and reduced verbal fluency could be of predictive value.

DISCUSSION

The current results support multilevel alterations of the language system in persons with schizophrenia. Creative expressions of psychotic experiences are frequently found but are not in the focus of this review. Clinical examinations of linguistic alterations can support differential diagnostics and early detection. Computational methods (Natural Language Processing) may improve the precision of corresponding diagnostics. The relations between language-related and other symptoms can improve diagnostics.

Trajectories through semantic spaces in schizophrenia and the relationship to ripple bursts

Human cognition is underpinned by structured internal representations that encode relationships between entities in the world (cognitive maps). Clinical features of schizophrenia-from thought disorder to delusions-are proposed to reflect disorganization in such conceptual representations. Schizophrenia is also linked to abnormalities in neural processes that support cognitive map representations, including hippocampal replay and high-frequency ripple oscillations. Here, we report a computational assay of semantically guided conceptual sampling and exploit this to test a hypothesis that people with schizophrenia (PScz) exhibit abnormalities in semantically guided cognition that relate to hippocampal replay and ripples. Fifty-two participants [26 PScz (13 unmedicated) and 26 age-, gender-, and intelligence quotient (IQ)-matched nonclinical controls] completed a category- and letter-verbal fluency task, followed by a magnetoencephalography (MEG) scan involving a separate sequence-learning task. We used a pretrained word embedding model of semantic similarity, coupled to a computational model of word selection, to quantify the degree to which each participant's verbal behavior was guided by semantic similarity. Using MEG, we indexed neural replay and ripple power in a post-task rest session. Across all participants, word selection was strongly influenced by semantic similarity. The strength of this influence showed sensitivity to task demands (category > letter fluency) and predicted performance. In line with our hypothesis, the influence of semantic similarity on behavior was reduced in schizophrenia relative to controls, predicted negative psychotic symptoms, and correlated with an MEG signature of hippocampal ripple power (but not replay). The findings bridge a gap between phenomenological and neurocomputational accounts of schizophrenia.

Natural language processing for mental health interventions: a systematic review and research framework

Neuropsychiatric disorders pose a high societal cost, but their treatment is hindered by lack of objective outcomes and fidelity metrics. AI technologies and specifically Natural Language Processing (NLP) have emerged as tools to study mental health interventions (MHI) at the level of their constituent conversations. However, NLP's potential to address clinical and research challenges remains unclear. We therefore conducted a pre-registered systematic review of NLP-MHI studies using PRISMA guidelines (osf.io/s52jh) to evaluate their models, clinical applications, and to identify biases and gaps. Candidate studies (n = 19,756), including peer-reviewed AI conference manuscripts, were collected up to January 2023 through PubMed, PsycINFO, Scopus, Google Scholar, and ArXiv. A total of 102 articles were included to investigate their computational characteristics (NLP algorithms, audio features, machine learning pipelines, outcome metrics), clinical characteristics (clinical ground truths, study samples, clinical focus), and limitations. Results indicate a rapid growth of NLP MHI studies since 2019, characterized by increased sample sizes and use of large language models. Digital health platforms were the largest providers of MHI data. Ground truth for supervised learning models was based on clinician ratings (n = 31), patient self-report (n = 29) and annotations by raters (n = 26). Text-based features contributed more to model accuracy than audio markers. Patients' clinical presentation (n = 34), response to intervention (n = 11), intervention monitoring (n = 20), providers' characteristics (n = 12), relational dynamics (n = 14), and data preparation (n = 4) were commonly investigated clinical categories. Limitations of reviewed studies included lack of linguistic diversity, limited reproducibility, and population bias. A research framework is developed and validated (NLPxMHI) to assist computational and clinical researchers in addressing the remaining gaps in applying NLP to MHI, with the goal of improving clinical utility, data access, and fairness.

Speech as a Graph: Developmental Perspectives on the Organization of Spoken Language

Language has been used as a privileged window to investigate mental processes. More recently, descriptions of psychopathological symptoms have been analyzed with the help of natural language processing tools. An example is the study of speech organization using graph theoretical approaches that began approximately 10 years ago. After its application in different areas, there is a need to better characterize what aspects can be associated with typical and atypical behavior throughout the lifespan, given the variables related to aging as well as biological and social contexts. The precise quantification of mental processes assessed through language may allow us to disentangle biological/social markers by looking at naturalistic protocols in different contexts. In this review, we discuss 10 years of studies in which word recurrence graphs were adopted to characterize the chain of thoughts expressed by individuals while producing discourse. Initially developed to understand formal thought disorder in the context of psychotic syndromes, this line of research has been expanded to understand the atypical development in different stages of psychosis and differential diagnosis (such as dementia) as well as the typical development of thought organization in school-age children/teenagers in naturalistic and school-based protocols. We comment on the effects of environmental factors, such as education and reading habits (in monolingual and bilingual contexts), in clinical and nonclinical populations at different developmental stages (from childhood to older adulthood, considering aging effects on cognition). Looking toward the future, there is an opportunity to use word recurrence graphs to address complex questions that consider biological/social factors within a developmental perspective in typical and atypical contexts.

Studying Psychosis Using Natural Language Generation: A Review of Emerging Opportunities

Disrupted language in psychotic disorders, such as schizophrenia, can manifest as false contents and formal deviations, often described as thought disorder. These features play a critical role in the social dysfunction associated with psychosis, but we continue to lack insights regarding how and why these symptoms develop. Natural language generation (NLG) is a field of computer science that focuses on generating human-like language for various applications. The theory that psychosis is related to the evolution of language in humans suggests that NLG systems that are sufficiently evolved to generate human-like language may also exhibit psychosis-like features. In this conceptual review, we propose using NLG systems that are at various stages of development as in silico tools to study linguistic features of psychosis. We argue that a program of in silico experimental research on the network architecture, function, learning rules, and training of NLG systems can help us understand better why thought disorder occurs in patients. This will allow us to gain a better understanding of the relationship between language and psychosis and potentially pave the way for new therapeutic approaches to address this vexing challenge.

Theory-Driven Analysis of Natural Language Processing Measures of Thought Disorder Using Generative Language Modeling

BACKGROUND

Natural language processing (NLP) holds promise to transform psychiatric research and practice. A pertinent example is the success of NLP in the automatic detection of speech disorganization in formal thought disorder (FTD). However, we lack an understanding of precisely what common NLP metrics measure and how they relate to theoretical accounts of FTD. We propose tackling these questions by using deep generative language models to simulate FTD-like narratives by perturbing computational parameters instantiating theory-based mechanisms of FTD.

METHODS

We simulated FTD-like narratives using Generative-Pretrained-Transformer-2 by either increasing word selection stochasticity or limiting the model's memory span. We then examined the sensitivity of common NLP measures of derailment (semantic distance between consecutive words or sentences) and tangentiality (how quickly meaning drifts away from the topic) in detecting and dissociating the 2 underlying impairments.

RESULTS

Both parameters led to narratives characterized by greater semantic distance between consecutive sentences. Conversely, semantic distance between words was increased by increasing stochasticity, but decreased by limiting memory span. An NLP measure of tangentiality was uniquely predicted by limited memory span. The effects of limited memory span were nonmonotonic in that forgetting the global context resulted in sentences that were semantically closer to their local, intermediate context. Finally, different methods for encoding the meaning of sentences varied dramatically in performance.

CONCLUSIONS

This work validates a simulation-based approach as a valuable tool for hypothesis generation and mechanistic analysis of NLP markers in psychiatry. To facilitate dissemination of this approach, we accompany the paper with a hands-on Python tutorial.

Linguistic correlates of suicidal ideation in youth at clinical high-risk for psychosis

Suicidal ideation (SI) is prevalent among individuals at clinical high-risk for psychosis (CHR). Natural language processing (NLP) provides an efficient method to identify linguistic markers of suicidality. Prior work has demonstrated that an increased use of "I", as well as words with semantic similarity to "anger", "sadness", "stress" and "lonely", are correlated with SI in other cohorts. The current project analyzes data collected in an SI supplement to an NIH R01 study of thought disorder and social cognition in CHR. This study is the first to use NLP analyses of spoken language to identify linguistic correlates of recent suicidal ideation among CHR individuals. The sample included 43 CHR individuals, 10 with recent suicidal ideation and 33 without, as measured by the Columbia-Suicide Severity Rating Scale, as well as 14 healthy volunteers without SI. NLP methods include part-of-speech (POS) tagging, a GoEmotions-trained BERT Model, and Zero-Shot Learning. As hypothesized, individuals at CHR for psychosis who endorsed recent SI utilized more words with semantic similarity to "anger" compared to those who did not. Words with semantic similarity to "stress", "loneliness", and "sadness" were not significantly different between the two CHR groups. Contrary to our hypotheses, CHR individuals with recent SI did not use the word "I" more than those without recent SI. As anger is not characteristic of CHR, findings have implications for the consideration of subthreshold anger-related sentiment in suicidal risk assessment. As NLP is scalable, findings suggest that language markers may improve suicide screening and prediction in this population.

Assessing coherence through linguistic connectives: Analysis of speech in patients with schizophrenia-spectrum disorders

BACKGROUND

Incoherent speech is a core diagnostic symptom of schizophrenia-spectrum disorders (SSD) that can be studied using semantic space models. Since linguistic connectives signal relations between words, they and their surrounding words might represent linguistic loci to detect unusual coherence in speech. Therefore, we investigated whether connectives' measures are useful to assess incoherent speech in SSD.

METHODS

Connectives and their surrounding words were extracted from transcripts of spontaneous speech of 50 SSD-patients and 50 control participants. Using word2vec, two different cosine similarities were calculated: those of connectives and their surrounding words (connectives-related similarity), and those of free-of-connectives words-chunks (non-connectives similarity). Differences between groups in proportion of five types of connectives were assessed using generalized logistic models, and connectives-related similarity was analyzed through non-parametric multivariate analysis of variance. These features were evaluated in classification tasks to differentiate between groups.

RESULTS

SSD-patients used less contingency (e.g., because) (p = .008) and multiclass connectives (e.g., as) (p < .001) than control participants. SSD-patients had higher minimum similarity of multiclass (adj-p = .04) and temporality connectives (e.g., after) (adj-p < .001), narrower similarity-range of expansion (e.g., and) (adj-p = .002) and multiclass connectives (adj-p = .04), and lower maximum similarity of expansion connectives (adj-p = .005). Using connectives' features alone, SSD-patients and controls could be distinguished with 85 % accuracy.

DISCUSSION

Our results show that SSD-speech can be distinguished from speech of control participants with high accuracy, based solely on connectives' features. We conclude that including connectives could strengthen computational models to categorize SSD.

Towards a temporospatial framework for measurements of disorganization in speech using semantic vectors

Incoherent speech in schizophrenia has long been described as the mind making "leaps" of large distances between thoughts and ideas. Such a view seems intuitive, and for almost two decades, attempts to operationalize these conceptual "leaps" in spoken word meanings have used language-based embedding spaces. An embedding space represents meaning of words as numerical vectors where a greater proximity between word vectors represents more shared meaning. However, there are limitations with word vector-based operationalizations of coherence which can limit their appeal and utility in clinical practice. First, the use of esoteric word embeddings can be conceptually hard to grasp, and this is complicated by several different operationalizations of incoherent speech. This problem can be overcome by a better visualization of methods. Second, temporal information from the act of speaking has been largely neglected since models have been built using written text, yet speech is spoken in real time. This issue can be resolved by leveraging time stamped transcripts of speech. Third, contextual information - namely the situation of where something is spoken - has often only been inferred and never explicitly modeled. Addressing this situational issue opens up new possibilities for models with increased temporal resolution and contextual relevance. In this paper, direct visualizations of semantic distances are used to enable the inspection of examples of incoherent speech. Some common operationalizations of incoherence are illustrated, and suggestions are made for how temporal and spatial contextual information can be integrated in future implementations of measures of incoherence.

Clues from caregiver emotional language usage highlight the link between putative social environment and the psychosis-risk syndrome

Familial emotional word usage has long been implicated in symptom progression in schizophrenia. However, few studies have examined caregiver emotional word usage prior to the onset of psychosis, among those with a clinical high-risk (CHR) syndrome. The current study examined emotional word usage in a sample of caregivers of CHR individuals (N = 37) and caregivers of healthy controls (N = 40) and links with clinical symptoms in CHR individuals. Caregivers completed a speech sample task in which they were asked to speak about the participant; speech samples were then transcribed and analyzed for general positive (e.g. good) and negative (e.g., worthless) emotional words as well as words expressing three specific negative emotions (i.e., anxiety, anger, and sadness) using Linguistic Inquiry and Word Count (LIWC). Findings indicated that (1) CHR caregivers used more negative and anxiety words compared to control caregivers; and (2) less positive word usage among CHR caregivers were related to more positive symptomatology among CHR individuals. These findings point toward the utility of automated language analysis in assessing the intersections between caregiver emotional language use and psychopathology.

Reflections on the nature of measurement in language-based automated assessments of patients' mental state and cognitive function

Modern advances in computational language processing methods have enabled new approaches to the measurement of mental processes. However, the field has primarily focused on model accuracy in predicting performance on a task or a diagnostic category. Instead the field should be more focused on determining which computational analyses align best with the targeted neurocognitive/psychological functions that we want to assess. In this paper we reflect on two decades of experience with the application of language-based assessment to patients' mental state and cognitive function by addressing the questions of what we are measuring, how it should be measured and why we are measuring the phenomena. We address the questions by advocating for a principled framework for aligning computational models to the constructs being assessed and the tasks being used, as well as defining how those constructs relate to patient clinical states. We further examine the assumptions that go into the computational models and the effects that model design decisions may have on the accuracy, bias and generalizability of models for assessing clinical states. Finally, we describe how this principled approach can further the goal of transitioning language-based computational assessments to part of clinical practice while gaining the trust of critical stakeholders.

Referential noun phrases distribute differently in Turkish speakers with schizophrenia

In all human languages, noun phrases (NPs) (e.g., 'a field', 'the woman with a book') are used to identify entities in discourse. Previous evidence has shown that the spontaneous speech of patients with schizophrenia (Sz) shows differences in the distribution of grammatically different types of NPs, which are in part specific to patients with formal thought disorder (FTD). Here we sought to provide the first evidence of related grammatical effects in a non-Indo-European language. Results from a picture description task in a sample of 16 Turkish speakers with FTD (+FTD), 15 without FTD (-FTD), and 27 controls revealed that relative to controls, people with Sz over-produced NPs that are 'bare' (in the sense of lacking any grammatical items such as the or a in English). The +FTD group generally showed stronger effects than -FTD, and used more pronouns and less NPs co-referring with previously mentioned NPs. In addition, the dynamic distribution of NP types over narrative time showed an effect of increased mean distance between definite NPs in -FTD relative to controls. In +FTD but no other group there was an unexpected random distribution of indefinite DPs. Incidence rates of referential anomalies increased from controls to the -FTD and +FTD groups. These findings further confirm that Sz is manifest through specific linguistic effects in the referential structure of meaning as mediated by grammar. They provide a linguistic baseline for neurocognitive models of FTD and help to define appropriate targets for the automatic extraction of linguistic features to classify psychotic speech.

Syntactic complexity of spoken language in the diagnosis of schizophrenia: A probabilistic Bayes network model

In the clinical linguistics of schizophrenia, syntactic complexity has received much attention. In this study, we address whether syntactic complexity deteriorates within the six months following the first episode of psychosis in those who develop a diagnosis of schizophrenia. We collected data from a cohort of twenty-six first-episode psychosis and 12 healthy control subjects using the Thought and Language Index interview in response to three pictures from the Thematic Apperception Test at first assessment and after six months (the time of consensus diagnosis). An automated labeling (part-of-speech tagging) for specific syntactic elements calculated large and granular syntactic complexity indices with a focus on clause complexity as a particular case from this spoken language data. Probabilistic reasoning leveraging the conditional independence properties of Bayes networks revealed that consensus diagnosis of schizophrenia predicted a decrease in nominal subjects per clause among individuals with first episode psychosis. From the entire sample, we estimate a 95.4 % probability that a 50 % decrease in mean nominal subjects per clause after six months is explained by the presence of first episode psychosis. Among those with psychosis, a 30 % decrease in this clause-complexity index after six months of experiencing the first episode predicted with 95 % probability a consensus diagnosis of schizophrenia, representing a conditional relationship between a longitudinal decrease in syntactic complexity and a diagnosis of schizophrenia. We conclude that an early drift towards linguistic disorganization/impoverishment of clause complexity-at the granular level of nominal subject per clause-is a distinctive feature of schizophrenia that decreases longitudinally, thus differentiating schizophrenia from other psychotic illnesses with shared phenomenology.

Linguistic and non-linguistic markers of disorganization in psychotic illness

BACKGROUND

Disorganization, presenting as impairment in thought, language and goal-directed behavior, is a core multidimensional syndrome of psychotic disorders. This study examined whether scalable computational measures of spoken language, and smartphone usage pattern, could serve as digital biomarkers of clinical disorganization symptoms.

METHODS

We examined in a longitudinal cohort of adults with a psychotic disorder, the associations between clinical measures of disorganization and computational measures of 1) spoken language derived from monthly, semi-structured, recorded clinical interviews; and 2) smartphone usage pattern derived via passive sensing technologies over the month prior to the interview. The language features included speech quantity, rate, fluency, and semantic regularity. The smartphone features included data missingness and phone usage during sleep time. The clinical measures consisted of the Positive and Negative Symptom Scale (PANSS) conceptual disorganization, difficulty in abstract thinking, and poor attention, items. Mixed linear regression analyses were used to estimate both fixed and random effects.

RESULTS

Greater severity of clinical symptoms of conceptual disorganization was associated with greater verbosity and more disfluent speech. Greater severity of conceptual disorganization was also associated with greater missingness of smartphone data, and greater smartphone usage during sleep time. While the observed associations were significant across the group, there was also significant variation between individuals.

CONCLUSIONS

The findings suggest that digital measures of speech disfluency may serve as scalable markers of conceptual disorganization. The findings warrant further investigation into the use of recorded interviews and passive sensing technologies to assist in the characterization and tracking of psychotic illness.

Clinical and computational speech measures are associated with social cognition in schizophrenia spectrum disorders

In this study, we compared three domains of social cognition (emotion processing, mentalizing, and attribution bias) to clinical and computational language measures in 63 participants with schizophrenia spectrum disorders. Based on the active inference model for discourse, we hypothesized that emotion processing and mentalizing, but not attribution bias, would be related to language disturbances. Clinical ratings for speech disturbance assessed disorganized and underproductive dimensions. Computational features included speech graph metrics, use of modal verbs, use of first-person pronouns, cosine similarity of adjacent utterances, and measures of sentiment; these were represented by four principal components. We found that higher clinical ratings for disorganized speech were predicted by greater impairments in both emotion processing and mentalizing, and that these relationships remained significant when accounting for demographic variables, overall psychosis symptoms, and verbal ability. Similarly, a computational speech component reflecting insular speech was consistently predicted by impairment in emotion processing. There were notable trends for computational speech components reflecting underproductive speech and decreased content-rich speech predicting mentalizing ability. Exploratory longitudinal analyses in a small subset of participants (n = 17) found that improvements in both emotion processing and mentalizing predicted improvements in disorganized speech. Attribution bias did not demonstrate strong relationships with language measures. Altogether, our findings are consistent with the active inference model of discourse and suggest greater emphasis on treatments that target social cognitive and language systems.

The SPEAK study rationale and design: A linguistic corpus-based approach to understanding thought disorder

AIM

Psychotic symptoms are typically measured using clinical ratings, but more objective and sensitive metrics are needed. Hence, we will assess thought disorder using the Research Domain Criteria (RDoC) heuristic for language production, and its recommended paradigm of "linguistic corpus-based analyses of language output". Positive thought disorder (e.g., tangentiality and derailment) can be assessed using word-embedding approaches that assess semantic coherence, whereas negative thought disorder (e.g., concreteness, poverty of speech) can be assessed using part-of-speech (POS) tagging to assess syntactic complexity. We aim to establish convergent validity of automated linguistic metrics with clinical ratings, assess normative demographic variance, determine cognitive and functional correlates, and replicate their predictive power for psychosis transition among at-risk youths.

METHODS

This study will assess language production in 450 English-speaking individuals in Australia and Canada, who have recent onset psychosis, are at clinical high risk (CHR) for psychosis, or who are healthy volunteers, all well-characterized for cognition, function and symptoms. Speech will be elicited using open-ended interviews. Audio files will be transcribed and preprocessed for automated natural language processing (NLP) analyses of coherence and complexity. Data analyses include canonical correlation, multivariate linear regression with regularization, and machine-learning classification of group status and psychosis outcome.

CONCLUSIONS

This prospective study aims to characterize language disturbance across stages of psychosis using computational approaches, including psychometric properties, normative variance and clinical correlates, important for biomarker development. SPEAK will create a large archive of language data available to other investigators, a rich resource for the field.

Speech disturbances in schizophrenia: Assessing cross-linguistic generalizability of NLP automated measures of coherence

INTRODUCTION

Language disorders - disorganized and incoherent speech in particular - are distinctive features of schizophrenia. Natural language processing (NLP) offers automated measures of incoherent speech as promising markers for schizophrenia. However, the scientific and clinical impact of NLP markers depends on their generalizability across contexts, samples, and languages, which we systematically assessed in the present study relying on a large, novel, cross-linguistic corpus.

METHODS

We collected a Danish (DK), German (GE), and Chinese (CH) cross-linguistic dataset involving transcripts from 187 participants with schizophrenia (111DK, 25GE, 51CH) and 200 matched controls (129DK, 29GE, 42CH) performing the Animated Triangles Task. Fourteen previously published NLP coherence measures were calculated, and between-groups differences and association with symptoms were tested for cross-linguistic generalizability.

RESULTS

One coherence measure, i.e. second-order coherence, robustly generalized across samples and languages. We found several language-specific effects, some of which partially replicated previous findings (lower coherence in German and Chinese patients), while others did not (higher coherence in Danish patients). We found several associations between symptoms and measures of coherence, but the effects were generally inconsistent across languages and rating scales.

CONCLUSIONS

Using a cumulative approach, we have shown that NLP findings of reduced semantic coherence in schizophrenia have limited generalizability across different languages, samples, and measures. We argue that several factors such as sociodemographic and clinical heterogeneity, cross-linguistic variation, and the different NLP measures reflecting different clinical aspects may be responsible for this variability. Future studies should take this variability into account in order to develop effective clinical applications targeting different patient populations.

Lower cohesion and altered first-person pronoun usage in the spoken life narratives of individuals with schizophrenia

Usage of computational tools to quantify language disturbances among individuals with psychosis is increasing, improving measurement efficiency and access to fine-grained constructs. However, few studies apply automated linguistic analysis to life narratives in this population. Such research could facilitate the measurement of psychosis-relevant constructs such as sense of agency, capacity to organize one's personal history, narrative richness, and perceptions of the roles that others play in one's life. Furthermore, research is needed to understand how narrative linguistic features relate to cognitive and social functioning. In the present study, individuals with schizophrenia (n = 32) and individuals without a psychotic disorder (n = 15) produced personal life narratives within the Indiana Psychiatric Illness Interview. Narratives were analyzed using the Coh-Metrix computational tool. Linguistic variables analyzed were indices of connections within causal and goal-driven speech (deep cohesion), unique word usage (lexical diversity), and pronoun usage. Individuals with schizophrenia compared to control participants produced narratives that were lower in deep cohesion, contained more first-person singular pronouns, and contained fewer first-person plural pronouns. Narratives did not significantly differ between groups in lexical diversity, third-person pronoun usage, or total word count. Cognitive-linguistic relationships emerged in the full sample, including significant correlations between greater working memory capacity and greater deep cohesion and lexical diversity. In the schizophrenia group, social problem-solving abilities did not correlate with linguistic variables but were associated with cognition. Findings highlight the relevance of psychotherapies which aim to promote recovery among individuals with psychosis through the construction of coherent life narratives and increasing agency and social connectedness.