The evolution of psychiatric neuromodulation reflects a century-long pursuit to precisely and safely modulate dysfunctional brain circuits for therapeutic benefit. The growing landscape of brain stimulation modalities, including transcranial magnetic stimulation, focused ultrasound, therapeutic lesioning and deep brain stimulation, offers diverse therapeutic options in psychiatry. Clinical studies support these interventions as safe and effective for treatment-resistant psychiatric disorders, yet the lack of direct comparative trials and consensus criteria across modalities remains a critical gap for clinical translation. Here we review converging evidence that neuromodulation tools engage shared neural networks that mediate symptom relief, highlighting opportunities for harmonized optimization of circuit-based therapeutics. We synthesize advantages and limitations of neuromodulation techniques to establish an evidence-based framework for selecting and combining modalities across psychiatric indications. Finally, we look to the future and discuss emerging questions on patient selection, personalization, technology development, real-world effectiveness and safety for cross-modality comparison and combination. Amid rapid advances in brain stimulation therapies, this Perspective defines priorities for shaping the future of mental health care. The evolution of psychiatric neuromodulation aims to precisely modulate dysfunctional brain circuits for therapeutic benefit yet lacks direct comparative trials and consensus criteria. This Perspective presents evidence that neuromodulation tools engage shared neural networks, proposing an evidence-based framework for optimizing circuit-based therapies and shaping future mental health care.
Objective neurometric markers that track depressive symptom severity across the continuum remain limited, particularly in untreated populations. Although electroencephalography offers promise, most prior studies are inconsistent, restricted to case–control discrimination or difficult to interpret mechanistically. These limitations raise the possibility that symptom-related information is not confined to focal electroencephalography features but instead distributed across cortical networks. Here we show that self-reported depressive symptom severity is associated with large-scale phase-resetting dynamics and can be characterized by an integrated index derived from phase-resetting rate modulation (PRRM), a measure of source-locked, time-lagged modulation of phase-resetting rate from source to target electrodes. In a resting-state cross-sectional cohort of 276 untreated adults (139 males and 137 females; aged 18–59 years) spanning a broad symptom range, including 116 with moderate-to-severe symptoms, an integrated index aggregating multiple individually weak but collectively informative PRRM-based features in the upper-alpha and low-gamma bands showed cross-validated predictability of symptom scores (r > 0.6) and achieved approximately 80% accuracy in discriminating moderate-to-severe from low-symptom individuals. These findings provide evidence for a physiologically interpretable neurometric signature and suggest that the PRRM framework may support future objective assessment and longitudinal monitoring of depressive symptom severity in untreated populations, although independent validation in clinically characterized cohorts is needed before clinical application. This research identifies a neurometric signature for depressive symptom severity using phase-resetting rate modulation in untreated adults. An integrated index demonstrates strong predictive accuracy, suggesting potential for objective assessment and monitoring, pending further validation in clinical settings.
In this World View, mental health researcher Lucas Murrins Marques draws on his research and personal experience of suicide loss within his family to argue that mental health systems are designed to manage illness and crisis but often fail to detect the gradual erosion of wellbeing that precedes them.
Many adolescents seek mental health advice from general-purpose artificial intelligence (AI) chatbots, despite limited evaluations against youth-relevant clinical standards. Here we developed the Adolescent Mental Health AI Response Tool (AMHART), a clinician-informed instrument for auditing the presence of guideline-informed response content across seven domains, and applied it to four major AI chatbots. Candidate items were reviewed by clinicians for their evidence base, objectivity and clinical importance, yielding a subset with strong content validity and inter-rater reliability. We then configured a three-model large language model evaluator ensemble, assessed its agreement with human reference labels and scored multiturn conversations in which adolescent personas (varying by sex, diagnosis and symptom severity) engaged with ChatGPT-5, ChatGPT-4, Claude 4.5 and Gemini 2.5. AMHART scores indicated greater coverage of applicable criteria for anxiety than depression and for moderate than severe presentations. The findings suggest AMHART offers a reliable, clinician-informed benchmark for evaluating chatbot responses to adolescent anxiety and depression disclosures. The authors developed and tested the Adolescent Mental Health AI Response Tool (AMHART), a clinician-informed tool to evaluate AI chatbot responses to adolescents disclosing anxiety or depression. Applying AMHART to four widely used AI chatbots revealed that these models have greater coverage for anxiety than depression and for moderate than severe presentations, also identifying gaps for coping guidance.
Cognitive deficits are a core feature of first-episode psychosis (FEP) and are closely linked to impaired functioning. Higher cognitive reserve (CR) is a protective factor, whereas early developmental factors, such as lower birth weight (BW), may be associated with poorer cognition. This longitudinal cohort study examined whether BW was associated with 1-year functioning through CR and cognition in 117 patients with nonaffective FEP; 162 healthy controls contributed cross-sectional BW and CR comparisons. Serial mediation models tested verbal memory and working memory separately. BW did not differ between groups, but CR was lower in FEP (P < 0.001). Here we show that higher BW was associated with greater CR (β = 4.64, P = 0.014), which was associated with better verbal memory (β = 2.21, P = 0.001) and, in turn, better 1-year functioning (β = 0.07, P = 0.004). The direct BW-functioning effect was not statistically significant, supporting a specific indirect pathway through CR and verbal memory. Forte et al. investigated whether birth weight influences functional outcomes 1 year after first-episode psychosis, and whether this association is mediated by cognitive reserve and cognitive performance, using a longitudinal cohort and serial mediation analyses.
