Category: Statistics

JMIR Med Inform. 2026 Aug 6;14:e86963. doi: 10.2196/86963.

ABSTRACT

BACKGROUND: Noninvasive brain stimulation may alleviate social media addiction, but its efficacy requires accurate individual targeting and real-time brain monitoring. The neural mechanisms underlying the effects of social media use (SMU) remain unclear, limiting the development of interventions. Understanding how different levels of SMU modulate brain activity could guide personalized neuromodulation strategies.

OBJECTIVE: This study investigated a cohort of young adults using a multistage design, with concurrent electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) to examine the effects of SMU on brain activity. It aimed to characterize static and dynamic neural changes at baseline and after a standardized SMU task in individuals with different daily SMU durations.

METHODS: Participants were all male and were divided into a heavy social media users (HSMU) group and a light social media users (LSMU) group based on self-reported daily SMU duration. All participants underwent baseline fMRI scanning, followed by an EEG-fMRI session immediately after a 2-hour controlled SMU task. Analyses were performed on the static and dynamic amplitudes of low-frequency fluctuations (sALFF and dALFF), static and dynamic functional connectivity (sFC and dFC), and EEG microstates.

RESULTS: At baseline, compared with the LSMU group, the HSMU group showed lower dALFF variability in the middle frontal gyrus. After the immediate-effect task, the LSMU group exhibited increased sALFF in the temporal lobe and decreased sALFF in the middle and superior frontal gyri. The HSMU group showed increased sALFF in the middle temporal gyrus and decreased sALFF in the inferior temporal gyrus, superior parietal gyrus, and prefrontal cortex. Regarding dALFF variability, the LSMU group showed a decrease in the superior medial frontal gyrus, whereas the HSMU group showed a decrease in the middle frontal gyrus and an increase in the calcarine cortex. sFC analysis revealed increased connectivity across nearly all networks in the LSMU group. Conversely, the HSMU group showed reduced sFC between the visual and default mode networks. The HSMU group showed significantly shorter duration of microstate A, shorter duration and lower coverage of microstate C, and longer duration and higher coverage of microstate D.

CONCLUSIONS: This study is the first to characterize the distinct neural patterns associated with different levels of daily SMU, using both EEG and fMRI to assess sALFF and dALFF alterations, widespread functional connectivity changes, and EEG microstate reorganizations. These findings demonstrate the unique value of multimodal assessment in identifying potential neural targets for personalized neuromodulation in social media addiction. Future studies should explore whether modulating these identified neural markers can effectively alleviate addictive behaviors and improve clinical outcomes across diverse populations.

PMID:42561423 | DOI:10.2196/86963

Obstet Gynecol. 2026 Aug 7. doi: 10.1097/AOG.0000000000006396. Online ahead of print.

ABSTRACT

OBJECTIVE: To describe Pregnancy Mortality Surveillance System (PMSS)-identified underlying causes of pregnancy-related deaths and changes over time in the overall and cause-specific pregnancy-related mortality ratios in the United States, 2005-2024.

METHODS: For this descriptive study, all pregnancy-related deaths were identified in the Centers for Disease Control and Prevention’s PMSS from 2005 to 2024. The PMSS has nine mutually exclusive underlying cause-of-death categories: hemorrhage, infection or sepsis, amniotic fluid embolism, thrombotic pulmonary or other embolisms, hypertensive disorders of pregnancy, anesthesia complications, cerebrovascular accidents, cardiovascular conditions (including cardiomyopathy), and other noncardiovascular medical conditions. Beginning with 2020 data, a new cause-of-death code specific for coronavirus disease 2019 (COVID-19) was added to the infection or sepsis category. Cause-specific proportionate mortality was calculated overall and in 5-year intervals. In addition, trends in pregnancy-related mortality ratios, overall and by underlying cause of death, were assessed over time with Joinpoint regression. Joinpoint analysis identified statistically significant changes in trends (joinpoints) and in annual percent change (APC) across distinct time periods.

RESULTS: During 2005-2024, 14,310 pregnancy-related deaths were identified in PMSS. There was a significant increase in the overall pregnancy-related mortality ratio from 2018 to 2021 (APC 20.11, P<.001) and a significant decrease from 2021 to 2024 (APC -17.91, P<.001). Throughout the study period, the most prevalent causes of death were infection or sepsis, cardiovascular conditions (including cardiomyopathy), other noncardiovascular medical conditions, and hemorrhage. Cardiovascular conditions were the leading cause of death for 17 of the 20 years. Trends in cause-specific pregnancy-related mortality ratios differed by cause of death. Deaths resulting from infection or sepsis had the largest increase in 2018-2021 (APC 86.17, P<.001) and the largest decrease in 2021-2024 (APC -49.67, P<.001), which coincided with the timing of the COVID-19 pandemic. Hemorrhage deaths showed an increasing pregnancy-related mortality ratio from 2017 to 2024 (APC 6.84, P=.02). The pregnancy-related mortality ratio for other cardiovascular conditions decreased from 2020 to 2024 (APC -9.49, P=.04).

CONCLUSION: From 2005 to 2024, the most prevalent causes of pregnancy-related death were cardiovascular conditions, infection or sepsis, other noncardiovascular medical conditions, and hemorrhage. However, cause-specific trends in pregnancy-related mortality in the United States have changed over time. These differences underscore the importance of sustained surveillance to better inform clinical and public health efforts to address leading and emerging causes of pregnancy-related death.

PMID:42561415 | DOI:10.1097/AOG.0000000000006396

J Med Internet Res. 2026 Aug 6;28:e89963. doi: 10.2196/89963.

ABSTRACT

BACKGROUND: Large language models (LLMs) are increasingly applied in clinical decision support, yet their diagnostic performance in Chinese-language settings and under realistic clinical workflows remains unclear. In particular, how LLMs perform across diseases with different prevalence and under stepwise diagnostic processes has not been well characterized.

OBJECTIVE: This study aimed to evaluate the diagnostic capabilities of LLMs for common diseases and rare diseases using clinical vignettes within a hypothetico-deductive framework and to identify their potential and limitations for clinical diagnosis.

METHODS: We evaluated 4 Chinese LLMs (Doubao 1.5, DeepSeek-V3, Kimi K1.5, and Leftdoctor GPT 3.5) using 56 clinical cases (28 chronic obstructive pulmonary disease [COPD], and 28 relapsing polychondritis [RP]) sourced from the China Clinical Case Results Database (March 31-April 14, 2025). Patient information was provided incrementally, starting with the initial medical history, followed by physical examination, and laboratory results. Evaluation metrics included top-3 accuracy (RTop3D), top-1 accuracy (RTopD), final diagnostic accuracy (RFA), and mean reciprocal rank (MRR). Statistical analysis was performed using generalized estimating equations (GEE), Friedman tests, and Wilcoxon signed-rank tests with Bonferroni correction. In addition, a qualitative analysis was conducted to characterize recurrent patterns of diagnostic errors.

RESULTS: LLMs demonstrated significantly higher diagnostic accuracy for COPD compared to RP across all metrics (P<.001). Diagnostic accuracy improved after additional clinical information was provided, with the improvement mainly observed in RP cases. In RP, diagnostic accuracy increased from 32.14% to 71.43% for DeepSeek and from 35.71% to 78.57% for Doubao, whereas COPD accuracy remained consistently high across all diagnostic stages (82.14%-92.86%). For COPD, ranking performance was high and comparable among all models (MRR range: 0.82-0.89; P=.71). In RP, diagnostic performance differed significantly among models (MRR range: 0.10-0.39; P<.001). Qualitative analysis showed that COPD errors were mainly related to a failure to recognize specific features, whereas RP errors involved more diverse patterns, particularly the neglect of negative evidence and the failure to recognize specific features.

CONCLUSIONS: Chinese LLMs demonstrated relatively strong diagnostic performance for common diseases such as COPD, but lower and less stable performance for rare diseases such as RP. Additional clinical information improved diagnostic accuracy primarily in RP cases, although differences between models remained evident under diagnostically complex conditions. Error patterns in RP cases suggest that current LLMs remain limited in their ability to integrate complex clinical information and exclusionary findings. Careful evaluation and appropriate clinical oversight remain important for their application in clinical practice.

PMID:42561411 | DOI:10.2196/89963

JMIR Form Res. 2026 Aug 6;10:e92879. doi: 10.2196/92879.

ABSTRACT

BACKGROUND: Older adults living with dementia often have limited tolerance for intraoral procedures, which can restrict access to even basic oral health assessment in clinical care. Procedural burden-characterized by sustained intraoral manipulation, sensory overload, and limited opportunities for interruption-may represent a key barrier to equitable oral care. Intraoral scanning (IOS) enables stepwise, interruptible acquisition of oral data and may reduce this burden, but feasibility and tolerability in people living with dementia have not been systematically evaluated.

OBJECTIVE: This study aimed to assess the feasibility and tolerability of bedside maxillary IOS compared with a conventional impression procedure in people living with dementia.

METHODS: A single-center, nonrandomized, counterbalanced, crossover feasibility study was conducted in a geriatric ward. Nineteen hospitalized participants with dementia (mean age 83.5, SD 3.4 years; mean Mini-Mental State Examination score 18.2, SD 4.8) underwent both bedside maxillary IOS and a conventional alginate-based impression procedure during a single visit. The order of procedures was not randomized but assigned using an alternating allocation sequence based on chronological enrollment to counterbalance potential order effects. The primary outcome was feasibility, defined as procedure completion without premature termination due to gag reflex, defensive movements, participant refusal, or withdrawal of assent or cooperation. Secondary outcomes included participant-rated discomfort using a 0 to 10 visual analog scale (VAS); clinician-rated manageability using a 0 to 10 VAS, with higher scores indicating greater difficulty; adverse reactions, including gag reflex and defensive movements; and active bedside procedural time in minutes. Paired comparisons were performed using the Wilcoxon signed-rank tests for VAS outcomes, a sign test for active bedside procedural time, and McNemar tests for binary outcomes, with a 2-sided significance level of .05.

RESULTS: Participants reported significantly lower discomfort with IOS than with the conventional impression procedure (median 1.0, IQR 0.5-1.5 vs median 7.75, IQR 3.7-8.5; P<.001), and clinician-rated manageability was more favorable for IOS (median 2.0, IQR 1.0-3.0 vs median 5.0, IQR 3.0-6.5; P<.001). Active bedside procedural time was similar in the analysis of all initiated procedures, with a mean duration of 3.02 (SD 0.72) minutes for IOS and 2.92 (SD 0.78) minutes for the conventional impression procedure. Exploratory analyses showed no statistically detectable evidence of an order effect. Gag reflex occurred less frequently during IOS (1/19, 5.3% vs 9/19, 47.4%; P=.004).

CONCLUSIONS: Bedside IOS was feasible in hospitalized older adults living with dementia and was associated with lower immediate procedural burden than a conventional alginate-based impression procedure. These findings support further evaluation of IOS as a low-burden approach for oral data acquisition in dementia care, while downstream treatment effects and implementation in other settings require further study.

PMID:42561409 | DOI:10.2196/92879

JMIR Mhealth Uhealth. 2026 Aug 6;14:e83194. doi: 10.2196/83194.

ABSTRACT

BACKGROUND: Poststroke cognitive impairment (PSCI) is a common and disabling complication after stroke; however, early screening remains challenging due to limited access to neuropsychological testing and the high cost of neuroimaging. Portable, tablet-based eye-tracking technology may offer a scalable, low-cost solution for early PSCI detection.

OBJECTIVE: This study aimed to evaluate the clinical utility of a tablet-based, AI-driven eye-tracking system for early screening of PSCI at 3 months after acute ischemic stroke. We sought to quantify oculomotor-cognitive associations and develop a practical nomogram for individualized risk prediction.

METHODS: We prospectively enrolled 142 hospitalized patients with acute cerebral infarction between May 2023 and October 2024, of whom 122 completed the 3-month follow-up and were included in the final analysis, along with 20 healthy community-dwelling controls. All patients underwent tablet-based eye tracking (visual paired comparison and antisaccade tasks) during the acute phase, as well as baseline and 3-month neuropsychological assessments. PSCI was defined using validated cutoffs. Multivariable logistic regression was used to identify independent predictors, and a nomogram was constructed. Internal validation was performed using bootstrap resampling (1000 samples).

RESULTS: At 3 months, out of 122 patients, 47 (38.5%) met PSCI criteria. Compared with patients with non-PSCI (n=75), patients with PSCI showed significantly prolonged correct saccade latency (median 322.96, IQR 209.45-445.59 ms vs 194.55, IQR 141.50-299.75 ms; Z=-4.03, P<.001), increased uncorrected error rate (median 30.00%, IQR 15.00%-42.00% vs 5.00%, IQR 0.00%-28.00%; Z=-4.24, P<.001), and reduced novelty preference ratio (median 1.44, IQR 0.97-1.70 vs 2.12, IQR 1.27-4.56; Z=-3.44, P=.001). Multivariable analysis identified 4 independent predictors of 3-month PSCI: older age (odds ratio [OR] 1.067 per year, 95% CI 1.009-1.129; P=.02), lower education level (OR 0.841 per year, 95% CI 0.708-0.999; P=.049), higher NIHSS (National Institutes of Health Stroke Scale) scores (OR 1.557 per point, 95% CI 1.075-2.256; P=.02), and prolonged correct saccade latency (OR 1.004 per ms, 95% CI 1.000-1.007; P=.04). A nomogram incorporating these 4 factors achieved good discriminative performance (area under the receiver operating characteristic curve 0.86, 95% CI 0.793-0.927) with satisfactory calibration.

CONCLUSIONS: Age, education, admission NIHSS, and correct saccade latency were identified as possible independent predictors of 3-month PSCI in this cohort. The tablet-based eye-tracking system, when combined with clinical variables, may represent a feasible approach for early PSCI screening. A nomogram based on these variables demonstrated high accuracy and potential clinical utility for early PSCI identification. This approach may facilitate early identification of high-risk patients and enable timely, personalized interventions in resource-limited settings.

PMID:42561408 | DOI:10.2196/83194

JMIR Form Res. 2026 Aug 6;10:e86944. doi: 10.2196/86944.

ABSTRACT

BACKGROUND: Telemedicine offers promising solutions for improving access to care among older adults with chronic conditions, but there is limited evidence on how older patients navigate and engage with video telehealth in their home environments.

OBJECTIVE: This study aimed to assess older adults’ acceptability, usability, comfort, and engagement with pharmacist-led telemedicine visits conducted at home using behavioral observation separate from the telehealth platform.

METHODS: This descriptive mixed methods pilot study included a convenience sample of 20 older adult veterans (aged ≥65 years, ≥2 chronic conditions, and ≥5 medications) recruited from a Veterans Affairs (VA) health care system. A project manager observed and video-recorded participant preparation for and conduct during in-home telemedicine visits with a clinical pharmacist. Recordings were coded using structured protocols to assess verbal and nonverbal behavior, technological challenges, and environmental factors. Trained and reliable coders rated verbal and nonverbal comfort, frustration, and engagement during technology setup and the telemedicine appointment using Likert-type impression scales (intercoder reliability was good, Cohen κ=0.81). Video data were also transcribed and coded for behavioral events (eg, troubleshooting, verbal expressions of age or technology ability, and nonverbal adaptation) and analyzed using rapid qualitative analysis.

RESULTS: Participants expressed moderate engagement and comfort with technology overall. Verbal and nonverbal engagement significantly increased from setup to appointment (Z=2.03, P=.002). Technology-related challenges (eg, audio lag and troubleshooting) occurred in over half the visits but were often resolved through participant adaptation or support from the research staff. Participants displayed both frustration (eg, sighing and leaning away) and adaptability (eg, propping up tablets and retrieving medications). Verbal expressions reflected a mix of technology confidence and age-related limitations. Environmental distractions were present in some visits (eg, dogs barking or phones ringing), but also allowed for rich clinical engagement (eg, home tours and direct observation of medications).

CONCLUSIONS: In this small pilot sample of older veterans, participants generally remained engaged during in-home telemedicine visits despite technological and contextual barriers, suggesting that home-based telemedicine may be acceptable for some older adults when support is available. Addressing technical support, home environment considerations, and age-related perceptions of technology may increase telehealth usability and satisfaction among aging populations.

PMID:42561385 | DOI:10.2196/86944

JMIR Res Protoc. 2026 Aug 6;15:e97653. doi: 10.2196/97653.

ABSTRACT

BACKGROUND: Young adults aged 18 to 25 years in rural communities face barriers to sexual and reproductive health (SRH) information, including clinician shortages, clinic closures, and privacy concerns in close-knit communities. Many rely on online sources of varying quality; one analysis found that 40% of birth control video content is inaccurate or misleading. AI-enabled conversational health tools (chatbots) may provide scalable SRH information, but implementation may be constrained by low institutional trust, privacy concerns, and rural users’ underrepresentation in AI development. SARHAchat is an AI-enabled SRH conversational health tool developed through prior work. This protocol describes a study to co-design and evaluate SARHAchat with rural young adults, treating trust as a design input rather than a postdeployment outcome.

OBJECTIVE: This study aims to (1) identify multilevel determinants of trust and acceptability for AI-enabled conversational health tools among rural pregnancy-capable individuals aged 18 to 25 years in North and South Carolina; (2) co-design a SARHAchat prototype with rural stakeholders using human-centered design; and (3) evaluate its feasibility, acceptability, usability, trust, and implementation outcomes in a pilot. The goal is to develop methods that support responsible implementation of conversational AI health tools in rural and underserved communities.

METHODS: This protocol uses a 3-phase mixed methods design with exploratory sequential logic. In phase 1, we will conduct semistructured interviews (target n=24) with pregnancy-capable individuals aged 18 to 25 years living in Health Resources and Services Administration (HRSA)-designated rural counties in North and South Carolina to identify determinants of trust and acceptability and develop a conceptual framework. In phase 2, we will convene 5 to 8 stakeholders per session across 2 community feedback sessions to co-design and refine SARHAchat. In phase 3, we will conduct a nonrandomized mixed methods feasibility pilot with 75 pregnancy-capable individuals aged 18 to 25 years from rural counties, recruited through community-identified social media platforms.

RESULTS: This proposal was funded in February 2026. All 3 phases have been approved by the University of North Carolina at Chapel Hill Institutional Review Board (26-0669). A total of 14 participants have enrolled in phase 1. Phase 2 co-design activities are planned for August to October 2026, and phase 3 pilot recruitment is expected to begin in January 2027, with data collection concluding in October 2027. Findings are anticipated to be submitted for publication in April 2028.

CONCLUSIONS: This protocol describes a feasibility study of a community-informed AI-enabled conversational SRH tool for rural young adults. By identifying trust-related design needs, refining a prototype through co-design, and generating preliminary feasibility, acceptability, usability, and implementation data, the study aims to establish trust-by-design methods that guide future effectiveness testing and may transfer to other sensitive health domains and underserved communities.

PMID:42561383 | DOI:10.2196/97653

Ophthalmic Plast Reconstr Surg. 2026 Aug 6. doi: 10.1097/IOP.0000000000003246. Online ahead of print.

ABSTRACT

PURPOSE: The objective of this study is to investigate whether checkpoint inhibitor proteins, including programed death protein-1, programmed death-1-pathway ligand, cytotoxic T-lymphocyte associated protein-4, lymphocyte activation gene-3, and CD73, are implicated in orbitally invasive sinonasal undifferentiated carcinomas (SNUCs).

METHODS: Patients with orbitally invasive SNUC presenting to a single institution between 2020 and 2024 were identified. Age and gender match controls were identified. Immunohistochemical staining was performed for each of the checkpoint inhibitors. Using light microscopy, the number of positively staining cells per 40× field was recorded across 5 consecutive fields and averaged. The differences in expression between the 2 were compared via a Mann-Whitney analysis.

RESULTS: Six patients with orbitally invasive SNUC and 11 sinus mucosal controls were identified. Immunohistochemical analysis of these tumors demonstrated positivity in both SNUC specimen and normal sinus mucosa for all biomarkers tested (CD73, lymphocyte activation gene-3, cytotoxic T-lymphocyte associated protein-4, programmed death-1-pathway ligand, programed death protein-1). Expression of CD73 and programmed death-1-pathway ligand was statistically significantly higher in SNUC specimens compared with normal sinus controls (p = 0.0003 and p = 0.0111, respectively).

CONCLUSION: Specimens from patients with sinonasal undifferentiated carcinoma express increased levels of programmed death-1-pathway ligand and CD73 compared with sinus controls. The results discovered in this investigation represent a significant proof of principle that immunotherapy may be a promising approach to address a potentially devastating disease.

PMID:42561378 | DOI:10.1097/IOP.0000000000003246

Diabetes Care. 2026 Aug 6:dc261042. doi: 10.2337/dc26-1042. Online ahead of print.

ABSTRACT

OBJECTIVE: To investigate the link between continuous glucose monitoring (CGM) metrics and risk of severe hypoglycemia and diabetic ketoacidosis (DKA) in individuals of type 1 diabetes.

RESEARCH DESIGN AND METHODS: We pooled individual-level data from 10 studies of type 1 diabetes. From a 14-day baseline period, we derived CGM metrics (time in range [TIR] 70-180 mg/dL, time above range [TAR], time below range [TBR], sensor mean glucose, SD, and percent coefficient of variation [%CV]) and CGM-defined hypo- and hyperglycemic episodes. We used Cox regression to evaluate associations with severe hypoglycemia and diabetic ketoacidosis, with and without adjustment for HbA1c.

RESULTS: Among 1,550 participants, 65 severe hypoglycemia and 163 DKA events occurred over a median follow-up of 16 weeks. Higher TBR <70 and <54 mg/dL, SD, %CV, and CGM-defined hypoglycemic episodes were associated with greater risk of severe hypoglycemia independent of HbA1c (e.g., hazard ratio [HR] 1.52 [1.29-1.79] per 1-SD higher TBR <70). Lower TIR and higher mean glucose, SD, %CV, TAR, and hyperglycemic episodes were associated with higher DKA risk; these associations were attenuated after HbA1c adjustment, but TIR, sensor mean glucose, and SD remained statistically and clinically significant (e.g., HR 0.76, 95% CI 0.60-0.97 per 1-SD higher TIR).

CONCLUSIONS: CGM hypoglycemic and glycemic variability metrics were associated with severe hypoglycemia risk independent of HbA1c, whereas TIR, mean glucose, and SD provided information for DKA risk beyond HbA1c. This supports the complementary value of CGM and HbA1c in risk assessment for severe hypoglycemia and DKA.

PMID:42561376 | DOI:10.2337/dc26-1042

Crit Care Sci. 2026 Aug 3;38:e20260480. doi: 10.62675/2965-2774.20260480. eCollection 2026.

ABSTRACT

OBJECTIVE: We aimed to assess the effect of a short period of invasive mechanical ventilation after a successful spontaneous breathing trial, compared with immediate extubation, on the risk of reintubation within 48 hours.

METHODS: We conducted a systematic review and meta-analysis of randomized clinical trials comparing a short period of invasive mechanical ventilation following a successful spontaneous breathing trial versus immediate extubation. We searched PubMed®, Cochrane Central, Embase, Scopus, and Web of Science. Pairs of reviewers independently screened studies, extracted data, and assessed risk of bias with the Cochrane Risk of Bias tool 2.0. The primary outcome was reintubation within 48 hours after randomization. Data were pooled using a Hartung-Knapp-Sidik-Jonkman random-effects model. Trial sequential analysis was performed to determine whether the accumulated evidence was sufficient for definitive conclusions. Certainty of evidence was assessed using the GRADE approach. Secondary outcomes were reintubation or death within 7 days after extubation, intensive care unit and hospital length of stay, in-hospital mortality, and ventilator-free days up to day 28.

RESULTS: Our search identified 1,473 unique records, of which 4 randomized clinical trials (n = 1,071 patients) were eligible. The pooled risk ratio for reintubation within 48 hours in patients receiving a short period of invasive mechanical ventilation after a successful spontaneous breathing trial, compared with immediate extubation, was 0.48 (95% confidence interval 0.22 – 1.07; p = 0.06; I2 = 41.6%). Trial sequential analysis confirmed that the current evidence base is underpowered for definitive conclusions. The evidence was rated as low due to serious inconsistency and imprecision. No significant differences were observed for any of the secondary outcomes.

CONCLUSIONS: Among critically ill adults, a short period of invasive mechanical ventilation after a successful spontaneous breathing trial did not significantly reduce the risk of extubation failure within 48 hours compared with immediate extubation (low-certainty evidence). Further adequately powered trials are needed to clarify the clinical efficacy of this intervention.

PMID:42561371 | DOI:10.62675/2965-2774.20260480