Multiple Sclerosis and Related Disorders

Word-finding difficulty (tip-of-the-tongue phenomenon) is among the most prevalent changes reported by persons with multiple sclerosis (MS), but we lack validated performance metrics to assess this common symptom. Antonyms is a novel task evaluating retrieval of target phonological codes via a brief, clinically-feasible paradigm that controls for other cognitive processes. Test-retest reliability was good (0.85). Among 660 patients, Antonyms demonstrated better criterion validity than traditional language tasks (e.g., animal generation, Boston Naming Test). Demonstrating discriminate validity, Antonyms was specifically related to patient-reported difficulties with expressive language rather than working memory, executive/speed, or episodic memory. The Antonyms task is provided herein.

Background: Relapsing-remitting multiple sclerosis (RRMS) often progresses to secondary-progressive MS (SPMS), but diagnosis is typically retrospective, delaying potential interventions. Existing rule-based and machine learning approaches often lack prospective validation and provide no uncertainty estimates, limiting clinical adoption. Objective: To prospectively evaluate a conformal prediction (CP)-enabled random forest model for reproducing clinician-recorded disease course labels in the Swedish Multiple Sclerosis Registry (SMSReg) and compare its classifications with the Lorscheider criteria. Methods: The CP model was trained and calibrated on SMSReg-2022 (10,787 individuals, 85,461 hospital visits) and then validated on SMSReg-2025 (15,719 individuals, 45,077 hospital visits; not included in the SMSReg-2022 training/calibration/testing datasets). Performance was assessed at the hospital visit and individual levels and compared with the Lorscheider criteria. Results: The model maintained good calibration on the prospective dataset. At the per-hospital visit-level, the model accuracies were 92% (RRMS) and 95% (SPMS). At the individual level, sensitivity was 98%, specificity 89%, and F1-score 0.75. With clinician-recorded labels as reference, the model showed higher sensitivity than the Lorscheider criteria (98% vs. 71%), more transition assignments (90% vs. 57%), and higher F1-score (0.76 vs. 0.69). Conclusion: The CP-enabled model prospectively reproduced clinician-recorded disease course labels with calibrated uncertainty, supporting harmonization and decision support.

Introduction: Neuroinflammation is a known pathophysiologic mechanism in multiple sclerosis and a proposed mechanism in fibromyalgia and myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS). Positron Emission Tomography (PET) imaging of the translocator protein (TSPO) has been widely used to measure neuroinflammation, though second-generation ligands are highly sensitive to a single nucleotide polymorphism, rs6971, resulting in high, mixed, and low affinity binders. It is unknown whether this polymorphism is associated with neuroinflammatory disorders. Materials and Methods: Eighty-six total participant genotypes for TSPO binding affinity were identified through Sanger sequencing (30 fibromyalgia, 21 ME/CFS, 11 multiple sclerosis, and 24 healthy controls). Associations between patient group and genotype were assessed with Fishers Exact test. Associations between genotype and symptom severity from self-report questionnaires were assessed with Kruskal-Wallis tests. Results: In the present study individuals with fibromyalgia, (ME/CFS), or multiple sclerosis were more likely to have a mixed- or low-affinity genotypes compared to healthy controls. Presence of the low or mixed genotypes were also associated with greater self-reported fatigue severity and depressive symptoms. Discussion: These findings should be confirmed in larger samples but suggest functional consequences of rs6971 polymorphism that may be related to the pathophysiology of pain and fatigue illnesses.

Background. CNM-Au8, a catalytic nanocrystalline gold suspension, is proposed to support neuronal bioenergetics, but its effect on chronic lesion evolution in multiple sclerosis (MS) has not been reported. Objectives. To evaluate the effect of CNM-Au8 on chronic lesion tissue expansion (CLTE). Results. In this exploratory post hoc analysis of the VISIONARY randomised placebo-controlled trial, CLTE was quantified from serial FLAIR MRI. Fifty-five participants were included (CNM-Au8, n=40; placebo, n=15). Compared with placebo, CNM-Au8 was associated with lower lesion expansion rate (p=0.001; Cliffs =-0.58) and volume (p<0.001; Cliffs =-0.65). Conclusions. CNM-Au8 treatment was associated with substantially reduced chronic lesion tissue expansion.

Objective To examine whether prospectively recorded maternal BMI in early pregnancy is associated with offspring multiple sclerosis (MS) risk, including childhood-onset and adult-onset MS. Design Nationwide population-based cohort study. Setting Swedish national health and demographic registers. Participants 1,524,863 singleton children born in Sweden from 1992 to 2007 and followed through 2025. Exposures Maternal BMI recorded at the first antenatal visit, categorised as underweight (<18.5), normal weight (18.5-24.9), overweight (25.0-29.9), and obesity ([&ge;]30.0), and also modelled continuously. Main outcome measures MS in offspring, identified as at least three ICD-10 G35 codes recorded on separate dates. Cox models estimated adjusted hazard ratios (HR) and 95% confidence intervals (CI) adjusting for confounders. Results During follow-up, 1539 offspring (0.1%) were diagnosed with MS at a median age of 24 years (interquartile range, 20-27 years). MS incidence rates by maternal BMI category were 0.37 per 10,000 person-years for both underweight, and normal weight, 0.42 for overweight, and 0.45 for obesity. Compared with normal weight women, maternal overweight and obesity were associated with higher hazards of offspring MS, with adjusted HRs of 1.20 (95% CI, 1.05 to 1.37) and 1.35 (95% CI, 1.11 to 1.64), respectively. Onset-specific analyses showed similar findings for adult-onset MS and less precise but directionally consistent estimates for paediatric-onset MS. Each 1-unit increase in maternal BMI was associated with a 3%, 4%, and 3% higher hazard of overall, paediatric-onset, and adult-onset offspring MS, respectively. Conclusions In this population-based cohort study, higher maternal BMI were associated with an increased rate of MS in offspring. The findings provide large-scale prospective evidence that maternal adiposity may be related to offspring MS risk and support further investigation of the prenatal metabolic environment in the early-life origins of MS.

Background: Secondary progressive multiple sclerosis (SPMS) is traditionally viewed as predominantly neurodegenerative, with limited inflammatory activity. However, chronic active lesions characterised by persistent rim inflammation and gradual expansion may continue to contribute to tissue injury. Objectives: To characterise chronic lesion dynamics in SPMS and determine whether chronic lesion expansion contributes to tissue damage and neurodegeneration. Thirty-one patients with SPMS underwent longitudinal MRI including T1-weighted, FLAIR and diffusion imaging. White matter lesions were manually segmented and chronic lesion tissue expansion (CLTE) was quantified as annualised percentage change. Associations between lesion expansion, mean diffusivity (MD) changes and brain atrophy were assessed using correlation, regression and mixed-effects modelling. Results: CLTE remained prominent, with a mean annual expansion rate of 7.1 +- 4.1%. Expansion substantially exceeded the contribution of newly formed lesions and occurred at rates comparable to relapsing multiple sclerosis. CLTE demonstrated moderate intra-patient clustering (ICC = 0.23). MD increased significantly within lesions during follow-up and was associated with lesion expansion. Chronic lesion expansion was also strongly associated with ventricular enlargement (partial r = 0.72, p < 0.001). Conclusions: Chronic lesion expansion persists in SPMS and represents a major source of tissue injury associated with microstructural degeneration and brain atrophy.

Progressive multiple sclerosis (PMS) lacks blood-based biomarkers of the neurodegenerative processes driving disability. We measured synaptopodin, a postsynaptic protein, in neuronal-enriched extracellular vesicles (NEVs) across two independent cohorts: the SPRINT-MS trial (n=184) and a real-world validation cohort (n=268). A one SD increase in baseline synaptopodin was associated with ~32% lower risk of confirmed disability progression in PMS (p=0.011), with the strongest effect in secondary PMS (SPMS) (~57% lower odds, p=0.019). This SPMS association was replicated in the validation cohort (~60% lower odds, p=0.009). Higher baseline synaptopodin was associated with future slower whole-brain atrophy and preserved microstructural integrity across MRI metrics (p=0.007-0.013). Longitudinal changes in NEV synaptopodin were associated with disability worsening (p=0.007) and paralleled MRI microstructural changes. NEV synaptopodin, but not NfL, was associated with confirmed disability progression in PMS participants with both biomarkers measured (HR 0.71 vs 1.12, p=0.007 vs 0.22 for NEV synaptopodin and NfL, respectively) and SPMS (HR 0.58 vs 1.10, p=0.003 vs 0.50). Adding synaptopodin, but not NfL, to a base model improved discrimination for disability progression in SPMS in both cohorts (e.g. validation AUC 0.755 to 0.838, p=0.005). These findings identify NEV synaptopodin as a prognostic and longitudinal, blood biomarker of synaptic integrity in PMS, particularly SPMS, complementing NfL.

The glymphatic system is the waste clearance pathway in the central nervous system. Although its dysfunction has been associated with several neurological disorders, its role in multiple sclerosis (MS) remains poorly understood. This system is thought to be particularly active during sleep, yet the relationship between sleep metrics and MRI-derived markers of glymphatic activity has been scarcely explored. In this study, we aimed to perform a multimodal magnetic resonance imaging (MRI) assessment of glymphatic function in MS and investigate its association with sleep parameters, clinical disability, disease duration and MRI-derived structural markers. The study included 22 patients (34.3{+/-}10.1 years; 16 female) with relapsing-remitting MS and 26 healthy controls (32.2{+/-}5.7 years; 18 female), all of whom underwent 3.0 Tesla MRI. Glymphatic function was assessed using diffusion tensor image analysis along the perivascular space (DTI-ALPS) index and global blood-oxygen-level-dependent (BOLD) signal-cerebrospinal fluid (CSF) coupling. In addition, sleep micro- and macrostructural features, clinical disability, disease duration, along with mean diffusivity (MD), lateral ventricular, choroid plexus volumes and lesion load were evaluated. Patients showed reduced DTI-ALPS index, though this difference became only a trend after controlling for MD. No significant group differences were observed in global BOLD signal-CSF coupling. However, in patients, the weaker coupling was associated with longer total sleep time, lower arousal index, greater disability and lateral ventricle enlargement, whereas the DTI-ALPS index showed no significant correlations with any of the measures. Together, these findings challenge the assumption that DTI-ALPS can serve as a reliable marker of glymphatic function in demyelinating diseases such as MS, where disease-related alterations in white matter microstructure can confound diffusion-based measures. In contrast, gBOLD-CSF coupling showed associations with sleep, clinical disability and structural changes in patients, despite the absence of a significant group-level difference. This finding suggests that gBOLD-CSF coupling may capture individual variability in glymphatic-related dynamics relevant to sleep and the clinical expression of MS.

Transcriptional risk scores, an integrated score of genomic susceptibility variants with expression Quantitative Trait Loci (eQTL) data, may more accurately distinguish clinical phenotypes compared to polygenic risk scores. This approach has not been applied to multiple sclerosis. We integrated data from the largest GWAS of MS, involving 14,802 MS patients, with eQTL data from 7,466 blood donors and 5,494 brain donors, using summary-data-based Mendelian randomization (SMR) followed by the heterogeneity in dependent instruments (HEIDI) test. MS SNVs with a significant eQTL were integrated with single-cell RNA sequencing data from 1,27 million peripheral blood mononuclear cells from 927 donors and 750,614 single nucleus RNA sequencing central nervous system cells from 192 individuals using Bayesian colocalization. Derived polygenic and cell-line specific transcriptional risk scores were compared with age-related MS severity score (ARMSS) in a cohort of 1,077 MS patients from the South Wales MS Registry and with MRI-derived age-adjusted total brain volumes of 272 MS patients within the UK Biobank. Cell-line specific transcriptional risk scores were associated with MS outcomes in two independent cohorts. Our results identified MS-associated eGenes enriched across several biological pathways, providing insights into pathophysiological mechanisms in MS that could facilitate precision medicine and identify novel druggable pathways.

Axonal degeneration is a principal driver of irreversible neurological disability in multiple sclerosis (MS), yet current treatments fail to target this neurodegenerative phase directly. Sterile alpha and TIR domain-containing protein 1 (SARM1) has emerged as an executioner of programmed axon destruction and is overexpressed in grey and white matter MS tissue, making its inhibition a promising therapeutic strategy. However, the effects of SARM1 inhibition on localized neurodegeneration versus systemic and central inflammation remain poorly understood in complex autoimmune environments like MS. Here, we investigated SARM1 pathology using global SARM1-/- knockout (SARM1-/-) mice, AAV-mediated CRISPR knockdown and overexpression of SARM1 in retinal ganglion cells (RGCs), and the small-molecule SARM1 inhibitor 5-iodoisoquinoline (5IIQ), across optic nerve crush (ONC) and experimental autoimmune encephalomyelitis (EAE) models. While global SARM1-/- did not alter the overall clinical course of EAE, it revealed a complex phenotype characterized by an altered peripheral inflammatory cytokine profile and persistent CNS immune infiltration, alongside preserved axonal and myelin integrity. RGC-restricted SARM1 knockdown partially preserved RGCs and axons during EAE, whereas SARM1 overexpression worsened both retinal function and axonal injury. Pharmacological inhibition with 5IIQ preserved axonal integrity and restored visual function in both the ONC and EAE models, as confirmed by electrophysiology, and reduced serum neurofilament light chain (NfL) in EAE, without affecting demyelination. Together, these findings demonstrate that SARM1 inhibition uncouples axonal self-destruction from demyelination and gross neuroinflammation while providing robust structural and functional neuroprotection. This study establishes SARM1 as a viable target for neuroprotective co-therapies designed to complement existing immunomodulatory and remyelinating regimens in MS and related neurodegenerative disorders.

Background: Quantitative susceptibility mapping (QSM) exhibits phase cancellation between paramagnetic (e.g. iron) and diamagnetic (e.g. myelin/protein) sources, obscuring microstructural lesion heterogeneity in multiple sclerosis (MS). We evaluated whether biophysical {chi}-separation identifies clinically meaningful lesion phenotypes and whether reduced echo sampling preserves microstructural sensitivity. Methods: In this cross-sectional study of 183 people with multiple sclerosis (PwMS), we analyzed 3327 quality-controlled white matter lesions using 3T MRI with a five-echo gradient-recalled echo sequence (GRE5) and a subsampled two-echo dataset (GRE2). A hybrid framework combining unsupervised clustering on whole-lesion paramagnetic ({chi}para) and diamagnetic ({chi}dia) features with rule-based compartmental edge-core gradient assignment resolved MS lesion phenotypes. Intralesional microstructural properties were evaluated using linear mixed-effects models. Multivariable linear regressions assessed associations with physical disability (Expanded Disability Status Scale, EDSS; n = 158), cognitive processing speed (Symbol Digit Modalities Test, SDMT; n = 79), and disease duration (DD; n = 161), with sensitivity models adjusting for total lesion volume, count, or DD. Results: The framework identified four reproducible lesion phenotypes (Adjusted Rand Index = 0.924): Type 1 (22.7%; preserved {chi}dia), Type 2 (32.9%; high {chi}para without positive edge-core gradient), Type 3 (35.8%; attenuated {chi}dia), and Type 4 (8.6%; high {chi}para with a positive edge-core paramagnetic gradient). Higher EDSS was selectively associated with more negative signed {chi}dia within Type 4 lesions (whole: {beta} = -0.37, q = 0.020; core: {beta} = -0.37, q = 0.006; edge: {beta} = -0.34, q = 0.030) and Type 2 edge ({beta} = -0.33, q = 0.044). Type 4 core associations remained robust across sensitivity models adjusting for total lesion volume ({beta} = -0.35, q = 0.004), total lesion count ({beta} = -0.37, q = 0.008), and DD ({beta} = -0.32, q = 0.033). Abbreviated GRE2 attenuated Type 4 compartmental associations to non-significance (all q greater than 0.05; core: {beta} = -0.20, q = 0.215). Conclusions: Quantitative {chi}-separation identified four internally stable susceptibility-defined MS lesion phenotypes, revealing disability-associated tissue heterogeneity beyond total lesion burden. Higher EDSS was associated with more negative diamagnetic measures within Type 4 cores after adjustment for lesion burden, pointing toward complex intralesional non-myelin tissue remodeling or reactive astrogliosis. Two-echo subsampling may influence compartment-specific clinical associations. By uncoupling susceptibility sources, {chi}-separation provides a quantitative framework for characterizing disability-associated lesion heterogeneity.

Aim. To evaluate the accuracy, run-to-run consistency, and agreement of large language models on the Ukrainian Integrated Licensing Examination Krok 2 for the Specialty of Physical Therapy. Materials and Methods. Five cloud-based LLMs (ChatGPT-5.5, Claude 5 Sonnet, Gemini 3.5 Flash, Grok 4, DeepSeek-V3) were evaluated using 150 Ukrainian-language multiple-choice questions from the official Krok 2 database. Each model completed two independent runs with identical inputs. Accuracy (with 95% confidence intervals, Wilson method), run-to-run consistency, and agreement were assessed. Statistical analysis included Cohen's {kappa}, Cochran's Q and pairwise McNemar tests with Holm-Bonferroni correction. Results. All models achieved >80% accuracy in both runs, exceeding the passing threshold of [&ge;]64%. In the first run, accuracy ranged from 80.00% (DeepSeek-V3) to 93.33% (Gemini 3.5 Flash); in the second run, from 82.67% (Claude 5 Sonnet) to 88.67% (Gemini 3.5 Flash). In the first run significant differences were observed between Gemini 3.5 Flash and Grok 4 (93.33% vs 84.00%; p=0.0022) and DeepSeek-V3 (93.33% vs 80.00%; p=0.0001). There were no statistically significant differences between models in the second run. R2R consistency exceeded 90% for all models with the highest in ChatGPT-5.5 (98.67%) and lowest in DeepSeek-V3 (90.00%). Cohen's {kappa} ranged from 0.64 to 0.94, indicating substantial to almost perfect agreement. The stability of LLMs' response correctness across two runs ranged from 77.33% (DeepSeek-V3) to 88.00% (Gemini 3.5 Flash). Conclusions. All LLMs demonstrated high performance on Ukrainian-language Krok 2 examination tasks. Gemini 3.5 Flash showed the highest accuracy in both runs and response correctness stability while ChatGPT-5.5 exhibited the greatest response consistency and agreement. Although the accuracy and stability of LLM responses may be interrelated, they represent distinct aspects of model performance. This highlights the need for a comprehensive approach to evaluating LLMs. The findings indicate the potential of LLMs as an auxiliary tool in the professional practice of physical therapy specialists.

Purpose: Stimulability testing involves the evaluation of multiple aspects of a patient's behavior, but the results are most frequently consolidated into a single domain (i.e., "stimulable"). The current study performs preliminary content, construct, concurrent, and predictive validation for the Multidimensional Stimulability Profile for Voice (MSP-V), a rating scale designed to measure multiple domains during stimulability testing (Degree of Change, Rapidity of Change, Generalization, Patient-rated Voice Quality, Patient-rated Voice Feel, and an overall rating of Stimulability Level). Methods: 78 patients with voice disorders (24 functional, 15 structural, 39 neurological) were administered the MSP-V by five speech-language pathologists (SLPs). Content validity was evaluated via the Content Validity Index (CVI) completed by nine voice-specialized SLPs. Construct validity was evaluated via exploratory factor analysis. Concurrent validity was assessed by determining whether MSP-V domains scaled in the expected direction compared to categorical stimulability ratings (e.g., "Highly stimulable" vs. "Stimulable"). Predictive validity was assessed by examining the relationship between MSP-V domains and voice disorder diagnosis, return to voice therapy, and patient-reported practice outside of therapy. Results: The CVI reached the critical value for clinician-rated and patient-rated domains. Two distinct factors were identified during EFA, one consisting of clinician-rated domains and the other of patient-rated domains. All MSP-V domains showed the expected directional relationships as categorical stimulability ratings. No domains differed significantly between disorder etiologies. The Rapidity of Change and Stimulability Level domains both significantly predicted return for voice therapy, where participants who rated lower on each domain were more likely to return to for voice therapy. No domains were predictive of patient-reported practice. Conclusions: The MSP-V is a clinically feasible tool with emerging content, construct, concurrent, and predictive validity that could be used to provide increased clarity and precision in stimulability testing documentation for research and practice.

Background: Temporomandibular disorders (TMDs) are common conditions in the general population. Several therapeutic approaches are available, but there is no consensus on the most effective, particularly in physiotherapy. Objectives: To assess the efficacy of manual therapy (MT) and therapeutic exercise (TE) in reducing pain in patients with TMDs. Methods: Four databases (MEDLINE, PEDro, EMBASE, CINAHL) were searched up to March 2025. We included randomized controlled trials (RCTs) evaluating the effects of MT and/or TE in individuals with TMDs, compared to any other rehabilitation treatment, with pain intensity as the primary outcome. The methodological quality was assessed using the PEDro scale. A network meta-analysis was performed to compare the effectiveness of different rehabilitative approaches. Results: 25 RCTs for a total of 1,540 people were included. The main comparisons were: MT and/or TE vs no treatment or placebo, occlusal splint, and instrumental physiotherapy. In most studies, MT and TE were more effective than controls in reducing pain. However, the network meta-analysis indicated that multimodal treatments were the most effective. The methodological quality of the included studies ranged from 3 to 9 points. Conclusion: MT and TE are effective strategies for reducing pain in subjects with TMDs and should be used as first-line interventions before more invasive treatments; the combination of dry needling and manual therapy appears to be the most effective approach. Nonetheless, the variability in study quality suggests that further high-quality research is needed to confirm these findings.

Background: Depression symptoms following traumatic brain injury (TBI) are common and strongly associated with adverse overall outcomes in military service members, veterans, and civilians. Cognitive behavioral therapy for depression (CBT-D) is considered a first-line treatment for depression symptoms in many contexts. However, costs associated with CBT-D can be substantial, and there are insufficient numbers of providers trained to perform CBT-D. In contradistinction, digital therapeutic interventions such as smartphone apps have virtually unlimited availability and low per-patient costs. However, evidence for efficacy of digital interventions providing CBT-D in the context of TBI is limited. The aims of this study were to design, develop, and test a novel smartphone app to provide CBT-D intended for military service members with depression symptoms and history of mild TBI. Results: We designed a novel smartphone app featuring digital avatar coaches to provide CBT-D intended for military service members with depression symptoms and history of mild TBI. Key features and goals of the 12 weekly CBT lessons included stigma minimization, personal engagement, and 1-2 hours per week of personal challenge homework assignments. We performed a randomized controlled trial initially involving military service members and veterans. Inclusion criteria were later relaxed to allow for civilian participation due to practical constraints . The control group received an app providing educational lessons about concussions and depression which did not include the therapeutic components of CBT. There was no difference between the groups randomized to CBT-D (n=19) vs. educational app control (n=23) in self-reported depression symptoms. Nevertheless, there was a statistically significant benefit of the CBT-D intervention in self-reported insomnia severity at the 16-week follow-up (four weeks post intervention) assessment (Cohens D=0.45, p=0.0064). There were also statistically significant positive correlations (r= 0.46 to 0.76, p=0.046 to 0.001) between changes over time in depression symptoms and changes over time in both insomnia and PTSD symptoms in the CBT-D group. There were no such correlations in the control group. Participants reported higher therapeutic alliance (p=0.0001) and overall mobile application user rating (p=0.036) for the CBT-D app than for the control education app. More favorable therapeutic alliance and mobile application user ratings correlated significantly with greater improvements in depression symptoms in the CBT-D group (r=0.53 to 0.66, p=0.02 to 0.002). However, dropout was substantial (40/59 in the CBT-D group and 31/54 in the control group) and results may not be fully representative of the population at risk. Conclusions: There remains a major unmet need for widely available, low-cost, effective interventions for depression symptoms in the context of mild TBI. The challenges of designing and rigorously assessing the effects such interventions remain formidable. However, the lessons learned from this study and indications of benefit in a highly relevant secondary outcome measure (insomnia) suggest that remote app-based interventions remain therapeutically viable. Trial Registration: clinicaltrials.gov NCT05147506, submitted 20 July 2021, posted 7 December 2021.

Background: Patients increasingly consult free artificial intelligence (AI) chatbots about their symptoms, but these models may favour agreement with the patient over accuracy. In obstructive sleep apnoea (OSA), where most cases in the general population remain undiagnosed and diagnosis depends on referral, this could validate symptom denial. We tested whether consumer chatbots maintain guideline-concordant referral when a simulated patient minimises symptoms. Methods: Seven guideline-concordant and standardised OSA vignettes (apnoea-hypopnoea-index, AHI 8-52 events/h; body mass index 28-42 kg/m2) were presented to five free chatbots (GPT-5, Gemini 2.5 Flash, Claude Sonnet 4.6, DeepSeek V4 Flash and Grok-4.3) under neutral and symptom-minimising conditions (10 simulated conversations per model-vignette-condition cell; N=700). Referral maintenance was scored by a structured rubric and every capitulation was human-confirmed. Results: All chatbots maintained referral recommendations in neutral conversations (350/350; 100%). When simulated patients minimised symptoms and resisted referral, maintenance fell to 64.3% (225/350), a reduction of 35.7 percentage points (95% CI 5.5-66.0; p=0.028). Rates ranged from 55.7% to 85.7% across chatbots, but evidence of between-chatbot differences was inconsistent across analyses. Rates differed across clinical vignettes (p=0.001). A variance analysis estimated greater variation between vignettes than between chatbots. In the three driving-risk vignettes, referral recommendations were abandoned in 46-68% of symptom-minimising conversations, pooling all five chatbots within each vignette. Conclusions: Consumer AI chatbots abandoned correct OSA referral recommendations in about one-third of symptom-minimising conversations, with higher observed abandonment in simulated cases involving driving risk. This may delay appropriate assessment and raises patient- and public-safety concerns.

Background Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and post-COVID condition are complex, debilitating illnesses characterized by persistent fatigue and post-exertional malaise. Recovery is uncommon without targeted intervention; however, emerging clinical reports suggest that a subset of patients experience full resolution of symptoms after engaging in mind-body approaches. Patient perspectives on these interventions are divided, and qualitative evidence exploring successful recovery pathways is rare. This protocol outlines a study to investigate the contextual factors and patient-constructed explanations surrounding full recovery from ME/CFS or post-COVID condition using mind-body approaches. Methods and analysis We will employ a qualitative descriptive design rooted in a constructivist paradigm. Purposive and snowball sampling will be used to recruit approximately 20 adults from established infection-associated fatigue networks who self-report full recovery (lasting three months or more) from ME/CFS or post-COVID condition following a mind-body approach. Data will be collected through semi-structured interviews conducted via Zoom, and investigator reflexivity will be reported. Demographic information, including comorbidities, will be documented to contextualize the sample. Data will be analyzed inductively using reflexive thematic analysis. To ensure research rigor and reporting transparency, the study will adhere to the 15-point checklist of criteria for good thematic analysis proposed by Braun and Clarke. Dissemination We will publish our findings in a peer-reviewed, open-access journal and present at relevant academic conferences. A lay summary of the results will also be shared directly with participating patient recovery networks.

Objective To determine the comparative effectiveness of different digital health interventions for adults with low back pain. Design Systematic review and network meta-analysis of randomised controlled trials. Data sources PubMed, Embase, the Cochrane Central Register of Controlled Trials (CENTRAL), CINAHL, APA PsycINFO, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), from inception to 19 March 2026 (updated to 8 August 2026). Study selection Randomised controlled trials enrolling adults with low back pain. Interventions were classified as telemedicine, web-based applications (Web-App), mobile applications (Mobile-App), virtual reality (VR), exergaming, biofeedback devices, face-to-face care, and control. Paired reviewers independently screened records, extracted data, and assessed risk of bias using the Cochrane Risk of Bias 2 tool. Methods Frequentist random-effects network meta-analyses synthesised pain intensity and physical function (primary outcomes), and pain-related fear avoidance, health-related quality of life, self-efficacy, anxiety, and depression (secondary outcomes), assessed at post-intervention (from the end of treatment to < 2 months post-intervention), short-term ([&ge;]2 to <6 months), mid-term ([&ge;]6 to <12 months), and long-term ([&ge;]12 months) follow-up. Effects were reported as standardised mean differences with 95% confidence intervals. Transitivity was examined across potential effect modifiers, and Confidence in Network Meta-Analysis (CINeMA) informed confidence ratings. Results Ninety-two reports representing 84 trials involving 11 649 participants were included. At post-intervention, compared with control, VR showed the largest estimated reduction in pain intensity (SMD -1.43, 95% CI -1.87 to -0.98; moderate confidence) and improvement in physical function (-0.86, -1.45 to -0.26; very low confidence). VR was also associated with lower pain-related fear avoidance (-1.14, -1.68 to -0.61) and better health-related quality of life (0.97, 0.57 to 1.36) (both moderate confidence). Exergaming, Mobile-App, Web-App and telemedicine were also associated with lower pain intensity, while Mobile-App, Web-App, telemedicine, and biofeedback devices were associated with better physical function. For self-efficacy, exergaming (0.77, 0.05 to 1.49) showed favourable post-intervention effects, although confidence in this estimate was very low. No intervention showed a clear effect on anxiety, with confidence ranging from very low to moderate across comparisons, whereas telemedicine was associated with lower depression (-0.22, -0.44 to 0.00; low confidence). Networks became progressively sparse at later follow-up, and evidence of sustained effects was inconsistent across interventions and outcomes. Global inconsistency was detected in the post-intervention pain-intensity network, with local inconsistency identified for selected exergaming comparisons, and substantial heterogeneity was present in several networks. Conclusions Among adults with low back pain, VR showed the largest estimated post-intervention effects versus control for pain intensity, physical function, pain-related fear avoidance, and health-related quality of life, although comparative effects differed across outcomes. Self-efficacy showed favourable estimates for selected interventions, whereas evidence for anxiety and depression was limited. Evidence beyond treatment completion was increasingly sparse, and confidence in many network estimates was low or very low; firm comparative recommendations therefore remain premature. Systematic review registration PROSPERO CRD420261339841

Background: Excessive inflammation in early phases post-ischemic stroke is associated with poor long-term outcome and increased stroke recurrence. Interventions capable of downregulating persistent states of inflammation could improve recovery and prevent secondary stroke. Cardiovascular exercise (CE) has well-established anti-inflammatory effects but its capacity to attenuate the exaggerated inflammatory response triggered by ischemic stroke is unclear. Methods: We investigated the impact of eight weeks of CE on the serum concentration of inflammatory mediators in the early subacute phase post-stroke. Seventy-one patients with firstever ischemic stroke were randomly assigned to moderate-to-vigorous CE (n=45) or standard care (SC) (n=26). Concentrations of CRP, IL-6, IL-8, IL-10, TNF- and IL-1{beta} were measured at baseline and after the interventions. Differences between groups were analyzed with linear mixed models adjusted for age and sex. Results: Compared with the CE group, the SC group showed significant increases in IL-6 ({beta}=0.585, 95% CI=0.164-1.006, p=0.0073), IL-8 ({beta}=0.160, 95% CI=0.004-0.317, p=0.0447), and TNF- ({beta}=0.0845, 95% CI=0.0137-0.1553, p=0.0206). The SC group also showed greater increases in CRP ({beta}=0.250, 95% CI=-0.407, 0.906, p=0.4499) and IL-10 ({beta}=0.251, 95% CI=-0.024, 0.527, p=0.0729) but group differences were not significant. IL-1{beta} levels remained stable and differences between groups were not significant ({beta}=-0.008, 95% CI=-0.058, 0.042, p=0.7618). Conclusions: When introduced in the early subacute phase of stroke recovery, moderate-tovigorous CE mitigates increases in inflammation, but the effect differs across mediators. To what extent reducing the inflammatory response in the early stages post-stroke with CE can impact longterm recovery and stroke recurrence needs to be determined.

Tissue-based assays provide broad screening for neural autoantibodies but remain vulnerable to technical and reader-dependent variability. We evaluated an integrated workflow combining automated 3,3'-diaminobenzidine tissue-based assay processing, digital whole-slide imaging, and interpretable deep learning. Model development used 2,357 retrospective serum and cerebrospinal fluid images referred from more than 80 institutions and processed in a centralized laboratory, with reference labels established by expert-consensus tissue-based assay interpretation. A prospective validation cohort included 200 images from 200 unique patients (154 positive and 46 negative; 100 serum and 100 cerebrospinal fluid). The same images were evaluated by two expert and two novice readers in a fixed-order two-round study, with unaided interpretation followed by artificial-intelligence-assisted interpretation after a washout of at least two weeks. The standalone binary classifier achieved an area under the receiver operating characteristic curve of 0.922, accuracy of 86.5%, sensitivity of 85.7%, and specificity of 89.1% in prospective validation. Pooled reader accuracy was 90.4% in the unaided round and 97.4% in the artificial-intelligence-assisted round (McNemar P < 0.001), with specificity of 82.6% and 99.5%, respectively. Novice-reader accuracy was 83.3% unaided and 96.5% with artificial-intelligence assistance (P < 0.001), approaching unaided expert accuracy of 97.5%. Among tissue-based-assay-positive cases, reader accuracy for cytoplasmic/nuclear antigen-associated versus neural surface antigen-associated staining-pattern subclassification was 88.8% unaided and 94.0% with artificial-intelligence assistance (P < 0.001). These findings support further multicentre, cross-platform evaluation of artificial-intelligence-assisted neural autoantibody tissue interpretation.