Speech connectedness predicts reading performance three months in advance: a longitudinal experiment.

Aiming to verify the predictive value of oral narrative structure on reading acquisition, we followed 253 children (first and second graders) during an entire school year, assessing oral narratives and reading performances in five sessions. Transcriptions of oral narratives were represented as word-recurrence graphs to measure connectedness attributes. Connectedness predicted performance in phonological awareness, reading comprehension, and word reading accuracy 3-4 months in advance.

Detecting at-risk mental states for psychosis (ARMS) using machine learning ensembles and facial features.

AIMS: Our study aimed to develop a machine learning ensemble to distinguish "at-risk mental states for psychosis" (ARMS) subjects from control individuals from the general population based on facial data extracted from video-recordings. METHODS: 58 non-help-seeking medication-naïve ARMS and 70 healthy subjects were screened from a general population sample. At-risk status was assessed with the Structured Interview for Prodromal Syndromes (SIPS), and "Subject's Overview" section was filmed (5-10 min). Several features were extracted, e.g., eye and mouth aspect ratio, Euler angles, coordinates from 51 facial landmarks. This elicited 649 facial features, which were further selected using Gradient Boosting Machines (AdaBoost combined with Random Forests). Data was split in 70/30 for training, and Monte Carlo cross validation was used. RESULTS: Final model reached 83 % of mean F1-score, and balanced accuracy of 85 %. Mean area under the curve for the receiver operator curve classifier was 93 %. Convergent validity testing showed that two features included in the model were significantly correlated with Avolition (SIPS N2 item) and expression of emotion (SIPS N3 item). CONCLUSION: Our model capitalized on short video-recordings from individuals recruited from the general population, effectively distinguishing between ARMS and controls. Results are encouraging for large-screening purposes in low-resource settings.

Gesticulation in individuals with at risk mental states for psychosis.

Nonverbal communication (NVC) is a complex behavior that involves different modalities that are impaired in the schizophrenia spectrum, including gesticulation. However, there are few studies that evaluate it in individuals with at-risk mental states (ARMS) for psychosis, mostly in developed countries. Given our prior findings of reduced movement during speech seen in Brazilian individuals with ARMS, we now aim to determine if this can be accounted for by reduced gesticulation behavior. Fifty-six medication-naïve ARMS and 64 healthy controls were filmed during speech tasks. The frequency of specifically coded gestures across four categories (and self-stimulatory behaviors) were compared between groups and tested for correlations with prodromal symptoms of the Structured Interview for Prodromal Syndromes (SIPS) and with the variables previously published. ARMS individuals showed a reduction in one gesture category, but it did not survive Bonferroni's correction. Gesture frequency was negatively correlated with prodromal symptoms and positively correlated with the variables of the amount of movement previously analyzed. The lack of significant differences between ARMS and control contradicts literature findings in other cultural context, in which a reduction is usually seen in at-risk individuals. However, gesture frequency might be a visual proxy of prodromal symptoms, and of other movement abnormalities. Results show the importance of analyzing NVC in ARMS and of considering different cultural and sociodemographic contexts in the search for markers of these states.

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.

How can computational tools help to understand language patterns in mental suffering considering social diversity.

The complex interaction between biological and social factors challenges measuring human behavior. Language has been a crucial source of information that mirrors inner processes like thoughts. The development of a novel computational strategy that helps to understand language needs to consider social factors that could also impact human behavior. Ten years ago, I developed a computational approach based on graph theory to measure structural aspects of the narrative's mental organization expressed in spontaneous oral reports. It was possible to measure the decrease in narrative graph connectedness associated with the schizophrenia diagnosis and negative symptoms severity. However, I was worried that the psychiatric field neglected factors from diverse social realities (such as poor access to education). Formal education impacts language by mastering grammar and syntax. Changes in language structure could be related to symptoms and lack of exposure to formal education. Indeed, the same connectedness markers increase according to typical cognitive and academic development. In this paper, I describe the reasons and methods for investigating both factors (psychiatric symptoms and formal education) on language patterns. Further, I evaluate concerns and future challenges of using computational strategies that include social diversity in mental health conditions.

Happy thoughts: What computational assessment of connectedness and emotional words can inform about early stages of psychosis.

In recent years, different natural language processing tools measured aspects related to narratives' structural, semantic, and emotional content. However, there is a need to better understand the limitations and effectiveness of speech elicitation protocols. The graph-theoretical analysis applied to short narratives reveals lower connectedness associated with negative symptoms even in the early stages of psychosis, but emotional topics seem more informative than others. We investigate the interaction between connectedness and emotional words with negative symptoms and educational level in participants with and without psychosis. For that purpose, we used a speech elicitation protocol based on three positive affective pictures and calculated the proportion of emotional words and connectedness measures in the first-episode psychosis (FEP) group (N: 24) and a control group (N: 33). First, we replicated the association between connectedness and negative symptoms (R2: 0.53, p: 0.0049). Second, the more positive terms, the more connected the narrative was, exclusively under psychosis and in association with education, pointing to an interaction between symptoms and formal education. Negative symptoms were independently associated with connectedness, but not with emotional words, although the associations with education were mutually dependent. Together, education and symptoms explained almost 70 % of connectedness variance (R2: 0.67, p < 0.0001), but not emotional expression. At this initial stage of psychosis, education seems to play an important role, diminishing the impact of negative symptoms on the narrative connectedness. Negative symptoms in FEP impact narrative connectedness in association with emotional expression, revealing aspects of social cognition through a short and innocuous protocol.

Ethical Implications of the Use of Language Analysis Technologies for the Diagnosis and Prediction of Psychiatric Disorders.

Recent developments in artificial intelligence technologies have come to a point where machine learning algorithms can infer mental status based on someone's photos and texts posted on social media. More than that, these algorithms are able to predict, with a reasonable degree of accuracy, future mental illness. They potentially represent an important advance in mental health care for preventive and early diagnosis initiatives, and for aiding professionals in the follow-up and prognosis of their patients. However, important issues call for major caution in the use of such technologies, namely, privacy and the stigma related to mental disorders. In this paper, we discuss the bioethical implications of using such technologies to diagnose and predict future mental illness, given the current scenario of swiftly growing technologies that analyze human language and the online availability of personal information given by social media. We also suggest future directions to be taken to minimize the misuse of such important technologies.

Imagetic and affective measures of memory reverberation diverge at sleep onset in association with theta rhythm.

The 'day residue' - the presence of waking memories into dreams - is a century-old concept that remains controversial in neuroscience. Even at the psychological level, it remains unclear how waking imagery cedes into dreams. Are visual and affective residues enhanced, modified, or erased at sleep onset? Are they linked, or dissociated? What are the neural correlates of these transformations? To address these questions we combined quantitative semantics, sleep EEG markers, visual stimulation, and multiple awakenings to investigate visual and affect residues in hypnagogic imagery at sleep onset. Healthy adults were repeatedly stimulated with an affective image, allowed to sleep and awoken seconds to minutes later, during waking (WK), N1 or N2 sleep stages. 'Image Residue' was objectively defined as the formal semantic similarity between oral reports describing the last image visualized before closing the eyes ('ground image'), and oral reports of subsequent visual imagery ('hypnagogic imagery). Similarly, 'Affect Residue' measured the proximity of affective valences between 'ground image' and 'hypnagogic imagery'. We then compared these grounded measures of two distinct aspects of the 'day residue', calculated within participants, to randomly generated values calculated across participants. The results show that Image Residue persisted throughout the transition to sleep, increasing during N1 in proportion to the time spent in this stage. In contrast, the Affect Residue was gradually neutralized as sleep progressed, decreasing in proportion to the time spent in N1 and reaching a minimum during N2. EEG power in the theta band (4.5-6.5 Hz) was inversely correlated with the Image Residue during N1. The results show that the visual and affective aspects of the 'day residue' in hypnagogic imagery diverge at sleep onset, possibly decoupling visual contents from strong negative emotions, in association with increased theta rhythm.

Graph structure analysis of speech production among second language learners of Spanish and Chinese.

Language experience shapes the gradual maturation of speech production in both native (L1) and second (L2) languages. Structural aspects like the connectedness of spontaneous narratives reveal this maturation progress in L1 acquisition and, as it does not rely on semantics, it could also reveal structural pattern changes during L2 acquisition. The current study tested whether L2 lexical retrieval associated with vocabulary knowledge could impact the global connectedness of narratives during the initial stages of L2 acquisition. Specifically, the study evaluated the relationship between graph structure (long-range recurrence or connectedness) and L2 learners' oral production in the L2 and L1. Seventy-nine college-aged students who were native speakers of English and had received classroom instruction in either L2-Spanish or L2-Chinese participated in this study. Three tasks were used: semantic fluency, phonemic fluency and picture description. Measures were operationalized as the number of words per minute in the case of the semantic and phonemic fluency tasks. Graph analysis was carried out for the picture description task using the computational tool SpeechGraphs to calculate connectedness. Results revealed significant positive correlations between connectedness in the picture description task and measures of speech production (number of correct responses per minute) in the phonemic and semantic fluency tasks. These correlations were only significant for the participants' L2- Spanish and Chinese. Results indicate that producing low connectedness narratives in L2 may be a marker of the initial stages of L2 oral development. These findings are consistent with the pattern reported in the early stages of L1 literacy. Future studies should further explore the interactions between graph structure and second language production proficiency, including more advanced stages of L2 learning and considering the role of cognitive abilities in this process.

A Brazilian bottom-up strategy to address mental health in a diverse population over a large territorial area - an inspiration for the use of digital mental health.

Brazil is a continental country with a history of massive immigration waves from around the world. Consequently, the Brazilian population is rich in ethnic, cultural, and religious diversity, but suffers from tremendous socioeconomic inequality. Brazil has a documented history of categorizing individuals with culturally specific behaviors as mentally ill, which has led to psychiatric institutionalization for reasons that were more social than clinical. To address this, a "network for psychosocial care" was created in Brazil, that included mental health clinics and community services distributed throughout the country. This generates local support for mental health rehabilitation, integrating psychiatric care, family support and education/work opportunities. These clinics and community services are tailored to provide care for each specific area, and are more attuned to regional culture, values and neighborhood infrastructure. Here we review existing reports about the Brazilian experience, including advances in public policy on mental health, and challenges posed by the large diversity to the psychosocial rehabilitation.  In addition, we show how new digital technologies in general, and computational speech analysis in particular, can contribute to unbiased assessments, resulting in decreased stigma and more effective diagnosis of the mental diseases, with methods that are free of gender, ethnic, or socioeconomic biases.

Nonsemantic word graphs of texts spanning ∼ 4500 years, including pre-literate Amerindian oral narratives.

Non-semantic word graphs obtained from oral reports are useful to describe cognitive decline in psychiatric conditions such as Schizophrenia, as well as education-related gains in discourse structure during typical development. Here we provide non-semantic word graph attributes of texts spanning approximately 4500 years of history, and pre-literate Amerindian oral narratives. The dataset assessed comprises 707 literary texts representative of 9 different Afro-Eurasian traditions (Syro-Mesopotamian, Egyptian, Hinduist, Persian, Judeo-Christian, Greek-Roman, Medieval, Modern and Contemporary), and Amerindian narratives (N = 39) obtained from a single ethnic group from South America (Kalapalo, N = 18), or from a mixed ethnic group from South, Central and North America (non-Kalapalo, N = 21). The present article provides detailed information about each text or narrative, including measurements of four graph attributes of interest: number of nodes (lexical diversity), repeated edges (short-range recurrence), largest strongly connected component (long-range recurrence), and average shortest path (graph length).

Low Speech Connectedness in Alzheimer's Disease is Associated with Poorer Semantic Memory Performance.

Connected speech is an everyday activity. We aimed to investigate whether connected speech can differentiate oral narrative production between adults with Alzheimer's disease (AD; n = 24) and cognitively healthy older adults (n = 48). We used graph attributes analysis to represent connected speech. Participants produced oral narratives and performed semantic, episodic, and working memory tasks. AD patients produced less connected narratives than cognitively healthy older adults. Connectedness was associated with semantic memory in AD and with episodic memory in controls. Word-graphs connectedness represents a practical tool to assess cognitive impairment in AD patients.

Network analysis of narrative discourse and attention-deficit hyperactivity symptoms in adults.

Previous research investigating language in attention-deficit hyperactivity disorder (ADHD) has demonstrated several deficits in many aspects. However, no previous study employed quantitative methodology providing objective measures that could be compared among different studies with diverse samples. To fill this gap, we used network analysis to investigate how ADHD symptomatology impacts narrative discourse, a complex linguistic task considered to be an ecological measure of language. Fifty-eight adults (34 females and 24 males) with a mean age of 26 years old and a mean of 17 years of educational level were administered the Adult Self-Rating Scale for ADHD symptomatology. They also completed the State-Trait Anxiety Inventory, the Beck Depression Inventory and the Urgency, Premeditation, Perseverance, Sensation Seeking Behavior Scale. Intelligence quotient was calculated. Individuals were asked to tell a story based on a wordless picture book. Speech was recorded and transcribed as an input to SpeechGraphs software. Parameters were total number of words (TNW), number of loops of one node (L1), repeated edges (RE), largest strongly connected component (LSC) and average shortest path (ASP). Verbosity was controlled. Statistical analysis was corrected for multiples comparisons and partial correlations were performed for confounding variables. After controlling for anxiety, depression, IQ, and impulsiveness ADHD symptomatology was positively correlated with L1 and negatively correlated with LSC. TNW was positively correlated with ADHD symptoms. In a subdomain analysis, both inattention and hyperactivity-impulsivity were negatively correlated with LSC. Only hyperactivity-impulsivity positively correlated with TNW and L1. Results indicated a correlation between ADHD symptoms and lower connectedness in narrative discourse (as indicated by higher L1 and lower LSC), as well as higher total number of words (TNW). Our results suggest that the higher the number of ADHD symptoms, the less connectivity among words, and a higher number of words in narrative discourse.

Lower speech connectedness linked to incidence of psychosis in people at clinical high risk.

BACKGROUND: Formal thought disorder is a cardinal feature of psychotic disorders, and is also evident in subtle forms before psychosis onset in individuals at clinical high-risk for psychosis (CHR-P). Assessing speech output or assessing expressive language with speech as the medium at this stage may be particularly useful in predicting later transition to psychosis. METHOD: Speech samples were acquired through administration of the Thought and Language Index (TLI) in 24 CHR-P participants, 16 people with first-episode psychosis (FEP) and 13 healthy controls. The CHR-P individuals were then followed clinically for a mean of 7 years (s.d. = 1.5) to determine if they transitioned to psychosis. Non-semantic speech graph analysis was used to assess the connectedness of transcribed speech in all groups. RESULTS: Speech was significantly more disconnected in the FEP group than in both healthy controls (p < .01) and the CHR-P group (p < .05). Results remained significant when IQ was included as a covariate. Significant correlations were found between speech connectedness measures and scores on the TLI, a manual assessment of formal thought disorder. In the CHR-P group, lower scores on two measures of speech connectedness were associated with subsequent transition to psychosis (8 transitions, 16 non-transitions; p < .05). CONCLUSION: These findings support the utility and validity of speech graph analysis methods in characterizing speech connectedness in the early phases of psychosis. This approach has the potential to be developed into an automated, objective and time-efficient way of stratifying individuals at CHR-P according to level of psychosis risk.

The History of Writing Reflects the Effects of Education on Discourse Structure: Implications for Literacy, Orality, Psychosis and the Axial Age.

BACKGROUND: Graph analysis detects psychosis and literacy acquisition. Bronze Age literature has been proposed to contain childish or psychotic features, which would only have matured during the Axial Age (∼800-200 BC), a putative boundary for contemporary mentality. METHOD: Graph analysis of literary texts spanning ∼4,500 years shows remarkable asymptotic changes over time. RESULTS: While lexical diversity, long-range recurrence and graph length increase away from randomness, short-range recurrence declines towards random levels. Bronze Age texts are structurally similar to oral reports from literate typical children and literate psychotic adults, but distinct from poetry, and from narratives by preliterate preschoolers or Amerindians. Text structure reconstitutes the "arrow-of-time", converging to educated adult levels at the Axial Age onset. CONCLUSION: The educational pathways of oral and literate traditions are structurally divergent, with a decreasing range of recurrence in the former, and an increasing range of recurrence in the latter. Education is seemingly the driving force underlying discourse maturation.

Lowering costs for large-scale screening in psychosis: a systematic review and meta-analysis of performance and value of information for speech-based psychiatric evaluation

Objective: Obstacles for computational tools in psychiatry include gathering robust evidence and keeping implementation costs reasonable. We report a systematic review of automated speech evaluation for the psychosis spectrum and analyze the value of information for a screening program in a healthcare system with a limited number of psychiatrists (Maputo, Mozambique). Methods: Original studies on speech analysis for forecasting of conversion in individuals at clinical high risk (CHR) for psychosis, diagnosis of manifested psychotic disorder, and first-episode psychosis (FEP) were included in this review. Studies addressing non-verbal components of speech (e.g., pitch, tone) were excluded. Results: Of 168 works identified, 28 original studies were included. Valuable speech features included direct measures (e.g., relative word counting) and mathematical embeddings (e.g.: word-to-vector, graphs). Accuracy estimates reported for schizophrenia diagnosis and CHR conversion ranged from 71 to 100% across studies. Studies used structured interviews, directed tasks, or prompted free speech. Directed-task protocols were faster while seemingly maintaining performance. The expected value of perfect information is USD 9.34 million. Imperfect tests would nevertheless yield high value. Conclusion: Accuracy for screening and diagnosis was high. Larger studies are needed to enhance precision of classificatory estimates. Automated analysis presents itself as a feasible, low-cost method which should be especially useful for regions in which the physician pool is insufficient to meet demand.

Dreaming during the Covid-19 pandemic: Computational assessment of dream reports reveals mental suffering related to fear of contagion.

The current global threat brought on by the Covid-19 pandemic has led to widespread social isolation, posing new challenges in dealing with metal suffering related to social distancing, and in quickly learning new social habits intended to prevent contagion. Neuroscience and psychology agree that dreaming helps people to cope with negative emotions and to learn from experience, but can dreaming effectively reveal mental suffering and changes in social behavior? To address this question, we applied natural language processing tools to study 239 dream reports by 67 individuals, made either before the Covid-19 outbreak or during the months of March and April, 2020, when lockdown was imposed in Brazil following the WHO's declaration of the pandemic. Pandemic dreams showed a higher proportion of anger and sadness words, and higher average semantic similarities to the terms "contamination" and "cleanness". These features seem to be associated with mental suffering linked to social isolation, as they explained 40% of the variance in the PANSS negative subscale related to socialization (p = 0.0088). These results corroborate the hypothesis that pandemic dreams reflect mental suffering, fear of contagion, and important changes in daily habits that directly impact socialization.

Lowering costs for large-scale screening in psychosis: a systematic review and meta-analysis of performance and value of information for speech-based psychiatric evaluation.

OBJECTIVE: Obstacles for computational tools in psychiatry include gathering robust evidence and keeping implementation costs reasonable. We report a systematic review of automated speech evaluation for the psychosis spectrum and analyze the value of information for a screening program in a healthcare system with a limited number of psychiatrists (Maputo, Mozambique). METHODS: Original studies on speech analysis for forecasting of conversion in individuals at clinical high risk (CHR) for psychosis, diagnosis of manifested psychotic disorder, and first-episode psychosis (FEP) were included in this review. Studies addressing non-verbal components of speech (e.g., pitch, tone) were excluded. RESULTS: Of 168 works identified, 28 original studies were included. Valuable speech features included direct measures (e.g., relative word counting) and mathematical embeddings (e.g.: word-to-vector, graphs). Accuracy estimates reported for schizophrenia diagnosis and CHR conversion ranged from 71 to 100% across studies. Studies used structured interviews, directed tasks, or prompted free speech. Directed-task protocols were faster while seemingly maintaining performance. The expected value of perfect information is USD 9.34 million. Imperfect tests would nevertheless yield high value. CONCLUSION: Accuracy for screening and diagnosis was high. Larger studies are needed to enhance precision of classificatory estimates. Automated analysis presents itself as a feasible, low-cost method which should be especially useful for regions in which the physician pool is insufficient to meet demand.