Sandra Anna Just

Introduction: The course of psychotic disorders typically involves relapses. Early warning signs vary between individuals and are difficult to detect in clinical practice, especially in outpatient settings. Speech provides a quantitative clinical marker for detecting such early warning signs. The EU Horizon project TRUSTING (A TRUSTworthy speech-based AI monitoring system for the prediction of relapse in individuals with schizophrenia) aims to develop and evaluate a speech-based monitoring system for predicting imminent psychotic relapses. The study will examine the potential for prospective relapse prediction, and feasibility and usability of the monitoring system. Methods and analysis: In this multicentre observational study, n=240 remitted and at-risk-of-relapse adults with psychotic disorders and a comparison group with n=120 healthy participants (matched by age and sex) will be examined at six sites and in six different languages (German, French, Dutch, English, Czech and Turkish). The follow-up period is 6 months. The TRUSTING smartphone app will be used to collect weekly voice recordings through speech tasks; information on medication adherence, substance use, mood, anxiety and sleep quality; and motor data from a tapping task. Primary endpoints encompass model performance for relapse prediction, user adherence, transcription quality, usability of recordings and overall system usability. The primary analysis of user adherence, transcription quality, usability of recordings and overall system usability will be an unadjusted description of the respective proportions using 95% Wilson confidence intervals. Regarding relapse prediction, the predictive value of the risk estimates for relapse occurrence will be assessed using the area under the receiver operating characteristic curve. Exploratory analysis will be performed on potential speech-based markers associated with relapse risk. Ethics and dissemination: This study has been approved by swissethics (Business Administration System for Ethics Committees number: 2025–01177). Findings from this project will be disseminated through peer-reviewed journal publications and presentations at relevant scientific conferences, as well as at public events related to mental health. Trial Registration: ClinicalTrials.gov ID: NCT07397975 .

The usefulness of automated speech and language markers to monitor or predict psychotic symptoms depends on their ability to detect changes in mental state. To date, research linking psychosis and Natural Language Processing (NLP) has been conducted almost exclusively using cross-sectional experimental designs, limiting insight into how intraindividual speech changes relate to symptoms and functioning. Therefore, here we examined the speech of 95 individuals diagnosed with schizophrenia ( n = 70) or schizoaffective disorder ( n = 25) in a longitudinal design. Speech samples (using the Narrative of Emotions Task) and clinical ratings (PANSS, MINI-ICF) were collected at baseline and again at six-month follow-up. We computed longitudinal changes in eight NLP markers reflecting speech quantity, lexical and syntactic complexity, semantic similarity and connectivity. Our results replicated established associations between NLP metrics and clinical symptoms at baseline, for example local semantic similarity between adjacent sentences was negatively correlated with the PANSS negative symptom scale. Longitudinal analyses showed that measures of moving average type-token ratio, total token count, semantic similarity and graph connectivity were most sensitive to intraindividual changes in symptom severity and functional impairment. Finally, partial least squares regression yielded a significant association between longitudinal changes in these metrics and a component reflecting change in total and negative symptoms. Overall, our findings support the potential of NLP methods to help monitor symptoms for individuals with psychotic disorders. By enabling the early detection of symptom worsening, NLP tools could enable more timely intervention to prevent more serious illness.

Language is central to psychosis, making it a key focus for both clinical evaluation and research. Many automated linguistic metrics have been developed to analyze speech in psychosis, yet few studies have compared them systematically. In this study, we benchmarked the performance and robustness of 51 natural language processing (NLP) metrics using two datasets: one with German- and one with Russian-speaking participants. Metric performance was defined as the metrics’ capacity to reflect symptom severity and distinguish individuals with psychosis from healthy controls. The best-performing metrics tended to be robust across different clinical rating scales and speech elicitation tasks. Overall, verbosity metrics (e.g., word count) outperformed most other metrics across tasks and languages. Thus, verbosity-based metrics should be used as a baseline when evaluating cross-linguistic NLP metrics that are both clinically meaningful and useful for psychosis research.

Background: Psychotic disorder represents a leading cause of disability worldwide, and relapse in psychosis is common. Artificial intelligence (AI) is increasingly recognized as a method that could aid clinical monitoring for individuals experiencing psychosis. Objective: This review aims to map the existing literature on AI-based approaches—including machine learning, deep learning, and natural language processing—used to detect relapse in individuals with psychotic disorders. Methods: A systematic search strategy was conducted on PubMed, PsycINFO, and Embase up to January 7, 2026. Observational studies, randomized controlled trials, and quasi-experimental studies that used AI methods to detect relapse in psychosis were eligible for inclusion. Screening and data extraction procedures were conducted by at least 2 reviewers working independently. Findings were extracted, charted, and described using narrative synthesis based on data extraction and consensus meetings with the research team. The scoping review was prospectively registered with the Open Science Framework. Results: Relevant studies identified (N=10) included the use of digital tools such as smartphone- and smartwatch-based monitoring, ecological momentary assessment tools, social media activity, and internet searches. Digital phenotyping via smartphones and wearables emerged as the most common method for data collection. The efficacy of AI models varied with sensitivity (or recall) ranging from 0.25 to 0.77 and specificity (or precision) ranging from 0.06 to 0.88. The reported area under the receiver operating characteristic curve for models ranged from 0.63 to 0.78. AI models were heterogeneous across studies, and most study findings were not replicated. Conclusions: This scoping review highlights both the promise and the current limitations of AI in psychosis relapse detection. Passive digital phenotyping research in the detection of psychosis relapse has progressed, and personalized approaches with individual-level modeling show promise; however, further studies need to include larger numbers of participants and should incorporate methods such as large language models. Future studies will require large collaborations aimed at delivering AI methods for use in real-world clinical practice.

Formal thought disorder (FTD) is a core psychosis feature. Disentangling its dimensions requires tasks simple enough for formal modeling yet sensitive enough to capture individual variation across the psychosis spectrum. The semantic verbal fluency task offers precisely this: a structured behavioral trace of semantic memory sampling, amenable to computational analysis using distributed word embeddings. We hypothesized that this sampling process is governed by two dissociable mechanisms mapping onto FTD dimensions: initial retrieval drive (d ₀ ), quantifying the motivational resource sustaining production, and semantic search precision (α), quantifying how strongly similarity to the preceding word constrains each retrieval step from near-random to highly structured. We hypothesized that reduced d ₀ would track negative psychosis symptoms and alogia, while degraded α would track language disorganization and left inferior longitudinal fasciculus (ILF) fractional anisotropy. We tested these predictions in a primary (N = 120) and an independent replication sample (N = 249) of German-speaking individuals across the psychosis spectrum. Both parameters decreased with greater psychosis severity and, in the primary sample, they dissociated regarding their clinical correlates. d ₀ correlated negatively with negative symptoms, general psychopathology, and poverty of speech, consistent with a computational signature of alogia. α correlated negatively with positive symptoms and cognitive flexibility, and, in individuals with psychosis, positively with left ILF fractional anisotropy. The association between d ₀ and negative symptoms was replicated in the independent sample. These findings pave the way for mechanistic, automatically derived FTD markers capturing subclinical variation across the psychosis spectrum and mapping onto underlying cognitive and neural processes.

Establishing whether observed behavioral differences reflect meaningful change in an individual necessitates baselines specific to the individual and task. Automated hybrid solutions combine adaptive baselines with fixed thresholds. Applying this approach to behavioral science harbors challenges: the pronounced gap between observable measurement and underlying construct means ground truth is typically unavailable. A stepwise framework is proposed to determine and evaluate the validity of baselines for longitudinal behavioral measurements.

Virtual reality (VR) relaxation offers an innovative, immersive approach to managing negative emotions. Such digital therapies represent a promising, growing field in mental health care but remain under-researched in older populations – a group in critical need of scalable, engaging treatments. We conducted a randomized controlled trial to evaluate the feasibility and effectiveness of VR-based relaxation compared to guided imagery (treatment-as-usual) in older adults with mental illness. 44 older psychiatric patients (aged 58–98) were randomized into either VR or guided imagery (GI). In total, 39 participants completed the study (VR: n = 21, GI: n = 18) and were selected for analysis. Dropout rates, satisfaction, and overall experience were indicators of feasibility in both groups. Additionally, motion sickness and sense of presence were assessed in the VR group. Effectiveness was evaluated through pre-post-measurements of state anxiety (STAI-X1), affect (PANAS), and visual analogue scales for stress, relaxation, and well-being. VR demonstrated strong feasibility, with low dropout rates, high satisfaction and immersion, and minimal motion sickness (M = 0.25 ± 0.91). Mixed repeated measures ANOVA analyses revealed significant improvements in state anxiety, stress, negative affect, relaxation, positive affect, and well-being across both groups (p < 0.001), with no significant differences between VR and GI. Feasibility and effectiveness were consistent across age, gender and severity of illness. Our findings suggest that VR relaxation is a feasible and effective intervention for older adults with mental illness, offering a comparable alternative to traditional relaxation methods. This study underscores the potential of VR to enhance mental health care for older populations, including those with advanced age and serious mental illness.

Language plays a central role in the assessment of individuals with psychosis, from taking a medical history to evaluating cognitive function. However, speaking multiple languages can significantly influence linguistic, cognitive and neural substrates. Therefore, it is essential to know whether an individual with psychosis is bilingual. Leaving modulating effects of bilingualism in psychosis unconsidered, runs high risk of confounding any clinical assessment and research. Although more than half of the world is bilingual, to date, this risk has not been addressed. This critical review challenges current basic diagnostic practices in psychiatry that conflate language and other cognitive domains. Drawing on neuropsychology, psycho-/neurolinguistics, and cognitive neuroscience, we (i) identify potential contact points between bilingualism and psychosis, (ii) present a decision tree framework for the clinical and research setting to systematically study those contact points, and (iii) provide the basis for developing and testing new treatments considering the lived realities of the majority of individuals with psychosis, namely bilingual individuals, and leveraging modern technology to do so. If the field of psychiatry embraces these conclusions, not only could bilingual individuals with psychosis experience more equity, but the larger field would benefit by reducing confounds inherent to ascribing to monolingual assumptions.

Care is central to medical practice. However, care is currently being re-defined due to the computational logic that is central to digital medicine. For those with serious mental illness in psychiatry, the new landscape of care has significant ethical, legal, and phenomenological implications. The obvious and less obvious differences in care between human caregivers and technology warrant reflection. This commentary examines the nature of care in digital medicine through a case study of voice-based chatbots used in the context of mental health. Several recommendations are derived to reconcile human-centered care with risks in digital medicine.