Category: Statistics

Traffic Inj Prev. 2026 May 5:1-11. doi: 10.1080/15389588.2026.2653672. Online ahead of print.

ABSTRACT

OBJECTIVE: This study aims to systematically investigate how key geometric parameters of spiral tunnels, specifically tunnel length and radius and travel direction, influence drivers saccadic eye movements and visual load.

METHODS: A field experiment was conducted using a wearable eye tracker to record saccadic behavior from 30 licensed drivers. Participants drove through 3spiral tunnels with varying lengths and radii under both uphill and downhill traversal conditions. Four saccade metrics (amplitude, duration, frequency, and velocity) were analyzed using descriptive statistics and ANOVA to evaluate visual workload. These metrics have been selected because they collectively reflect distinct aspects of visual scanning behavior: amplitude indicates the breadth of visual search, duration reflects the time required for processing fixated information, frequency represents the rate of gaze shifting, and velocity denotes the efficiency of oculomotor movement.

RESULTS: The findings indicate that tunnel geometry and travel direction significantly affect saccadic dynamics. Longer tunnels and smaller radii resulted in increased saccade amplitude, prolonged duration elevated frequency, and reduced velocity, suggesting heightened visual processing demand. Furthermore uphill traversal consistently produced larger amplitudes, longer durations higher frequencies, and slower velocities than downhill traversal across all tunnels, revealing a directional asymmetry in visual load.

CONCLUSIONS: This study demonstrates that spiral tunnel design, especially extended length and reduced radius, elevates drivers’ visual cognitive load with uphill travel imposing greater demands. The results provide empirical evidence to inform geometry-based design guidelines for optimizing visual ergonomics and improving operational safety in spiral tunnels.

PMID:42085709 | DOI:10.1080/15389588.2026.2653672

JMIR Med Educ. 2026 May 5;12:e84935. doi: 10.2196/84935.

ABSTRACT

BACKGROUND: Assessment is a critical component of teaching and learning and serves as the foundation for how learners demonstrate success in achieving learning objectives. Formative assessments (FAs) and timely feedback play a crucial role in integrated curricula, whereas basic and clinical sciences are taught in a coordinated manner. Feedback-based FA supports student learning, and teachers can determine learning gaps to monitor progress in learning. Based on existing evidence, limited literature compared the effect of online versus onsite FA on summative performance in a fully integrated curriculum.

OBJECTIVE: This study aimed to examine the effectiveness of online versus on-site FAs and feedback on summative assessment in the integrated medical curriculum.

METHODS: This study used an exploratory mixed methods approach to delving into students’ experiences with face-to-face versus online FA and feedback, and its effect on their summative performance in the integrated Bachelor of Medicine, Bachelor of Surgery program. This study was conducted at Fakeeh College for Medical Sciences in Jeddah, Saudi Arabia. A total of 143 consenting students were recruited into the study. The students in the study were distributed voluntarily into 2 groups regardless of age, sex, or academic performance. Group 1 (n=92) was assigned to receive online FAs and immediate online feedback throughout the module using the Speedwell system. However, Group 2 (n=51) was assigned to receive onsite FAs and face-to-face feedback throughout the module in the examination hall in the college. The quantitative part of the study involved analyzing student scores of summative assessments in 2 groups exposed to online and onsite FA and feedback. The qualitative part aimed to explore students’ perceptions of FA and feedback.

RESULTS: The passing rate in summative examinations (quiz, midmodule, and final) was higher in the onsite group (61.2%, 51%, and 62.7%, respectively) compared with the online group (53.3%, 48.3%, and 45.7%, respectively). However, the difference was statistically significant only in the quiz examination. Four key themes were identified from the qualitative analyses regarding participants’ different experiences of FA and feedback: the accessibility of the examination format facilitates flexibility in learning; FA is a means of recognizing learning opportunities; FAs help shift student attitudes toward learning; and the last theme is opportunities for discussion and personalized feedback.

CONCLUSIONS: This research sheds light on the intricate interplay between assessment modalities and student learning outcomes by demonstrating that onsite FA followed by onsite feedback is more effective than online FA and feedback in fostering student engagement and promoting deep understanding and improving students’ performance in summative examinations. Thereafter, this study contributes to the ongoing discourse surrounding effective assessment practices in contemporary educational settings.

PMID:42085703 | DOI:10.2196/84935

JMIR Res Protoc. 2026 May 5;15:e91726. doi: 10.2196/91726.

ABSTRACT

BACKGROUND: Tuberous sclerosis complex (TSC) is a rare genetic disorder caused by pathogenic variants in the TSC1 or TSC2 genes. Apart from multisystem physical manifestations, most individuals with TSC experience TSC-associated neuropsychiatric disorders (TAND). Little is known about how TAND severity changes over time and what factors may predict these changes. Preliminary data suggest the presence of differential TAND severity trajectories. Caregiver well-being may act as a mediator of TAND severity, and a well-being intervention designed for caregivers of children with developmental disabilities may improve caregiver well-being.

OBJECTIVE: The study aims are to (1) examine longitudinal trajectories of TAND severity in a large sample of individuals with TSC and to examine potential predictors of differential trajectories, (2) evaluate the association between caregiver well-being characteristics, TAND severity, and severity trajectories, and (3) adapt and evaluate the feasibility, acceptability, and potential efficacy of a brief, online group-based well-being intervention for family caregivers.

METHODS: For the first 2 aims, 500 individuals with TSC or their caregivers will be recruited in an accelerated longitudinal design to document TAND severity at 5 time points over 12 months via a web-based app. At each time point, participants will complete demographic, TSC characteristics, intervention, and well-being questionnaires. Data will be analyzed using latent class mixed and multinomial regression modeling (aim 1) and structural equation and mediation modeling (aim 2). Participatory methods will be used to adapt an existing caregiver well-being intervention for the TSC community (aim 3). Thirty caregivers will be invited to participate in the adapted group-based online well-being intervention.

RESULTS: This study was funded from July 2024 (HT94252410790 and HT94252410791), and ethics approvals were obtained from the University of Cape Town (July 2024), Vrije Universiteit Brussel (November 2024), and the Department of Defense Office of Human Research Oversight (December 2024). The TAND Toolkit app was adapted for longitudinal data collection (aims 1 and 2). Recruitment started in December 2025 and will continue until 500 participants are enrolled (anticipated December 2026). Primary outputs are expected by July 2028. For aim 3, experiential and adaptation workshops were completed in June 2025, the pilot intervention was delivered in November 2025, and data collection will continue till May 2026. Outputs are expected by December 2026.

CONCLUSIONS: Identification of differential longitudinal TAND trajectories and their correlates will stimulate research in TSC and generate evidence for the self-report quantified TAND checklist as a clinical outcome measure. Understanding the association between caregiver well-being and TAND severity will provide support for targeted well-being interventions. A successful pilot trial will provide preliminary data for larger-scale clinical trials, with the potential to support caregivers and improve TAND outcomes. Together, the findings from the study will help close the gap in interventions for TAND.

TRIAL REGISTRATION: ClinicalTrials.gov NCT06879665; https://clinicaltrials.gov/study/NCT06879665.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/91726.

PMID:42085697 | DOI:10.2196/91726

J Public Health Manag Pract. 2026 May 5. doi: 10.1097/PHH.0000000000002355. Online ahead of print.

ABSTRACT

OBJECTIVE: To assess the associations between sensitivity and specificity with prevalence in data matching.

METHODS: Using publicly available data, a synthetic dataset of names (“source population”; 8 million records) was created with records randomly assigned as positive or negative for a health outcome, as well as sex and birth date. All positives were included in file 1 (“disease registry”), and a random sample of positives and negatives were selected and merged to create file 2 (“study population”). The prevalence in the source population was defined as the proportion of individuals in the synthetic dataset who were randomly assigned as positive, and the prevalence in the study population as the proportion of individuals in the study population who were positive. Multiple disease registry and study population file pairs were created and matched with various prevalence in the source and study populations. Link Plus 3.0, a probabilistic record linkage program, was used for the data matching.

RESULTS: As the prevalence in the source population increases from 0.1% to 10%, the sensitivity increases from 80.0% to 94.6% and the specificity decreases slightly; as the prevalence in the study population increases from 10% to 99%, the sensitivity remains stable around 95.0% and the specificity stays at about 100.0%.

CONCLUSIONS: In data matching, the sensitivity is positively and the specificity is negatively associated with the prevalence in the source population, but not associated with the prevalence in the study population.

PMID:42085690 | DOI:10.1097/PHH.0000000000002355

Evolution. 2026 May 5:qpag080. doi: 10.1093/evolut/qpag080. Online ahead of print.

ABSTRACT

Adams and Collyer (2019) evaluated the statistical performance of several approaches for quantifying morphological modularity and found that EMMLi had inflated type I error rates and a bias towards more complex models compared to the Covariance Ratio (CR) approach. They suggested that this may have been at least partly driven by the fact that AICc values from EMMLi do not incorporate trait numbers, but this was not verified. Here I present a performance analysis of a trait-number corrected EMMLi approach (“EMMLip”), showing that this ameliorates rates of false discovery and produces conservative results that favor less complex models. The corrected EMMLi approach was effective at differentiating models of modularity with varying between- and within-module covariation especially when effect size or dataset size were sufficiently large. While CR tests remained more effective at specifically detecting overall modularity, I found that CR tests are sensitive to varying within/between module covariation, and in some cases had inflated model misspecification between 2- and 3-module hypotheses. With this minor correction (albeit incomplete), the combination of EMMLip and CR tests becomes the best available toolkit for detecting and contrasting modularity hypotheses. This toolkit is however still imperfect, and I discuss future avenues for improvements.

PMID:42085682 | DOI:10.1093/evolut/qpag080

JMIR Biomed Eng. 2026 May 5;11:e86624. doi: 10.2196/86624.

ABSTRACT

BACKGROUND: Breast cancer-related lymphedema (BCRL) is a chronic complication that impairs quality of life through persistent fluid accumulation. While clinical guidelines recommend longitudinal surveillance, implementation is often limited by the logistical challenges of frequent in-clinic visits. Bioelectrical impedance analysis (BIA), specifically the segmental extracellular water-to-total body water (ECW/TBW) ratio, offers a noninvasive method for tracking fluid status. However, the technical agreement between in-clinic and patient-led home-based BIA systems, as well as the feasibility of long-term self-monitoring in real-world settings, remains to be fully established.

OBJECTIVE: Our primary objective was to evaluate the agreement between in-clinic and home-based BIA systems for key body composition and fluid-related parameters. Our secondary objective was to characterize longitudinal fluid patterns and diurnal variations in ECW/TBW ratios to assess the feasibility of home-based monitoring.

METHODS: This prospective, patient-driven, 12-month observational study enrolled breast cancer survivors at risk for lymphedema. Agreement between the in-clinic BIA system (InBody 770) and the home-based device (BWA ON) was assessed using Bland-Altman analysis, intraclass correlation coefficients (ICCs), and the Lin concordance correlation coefficient (CCC). Longitudinal home-based ECW/TBW measurements were analyzed using linear mixed-effects models to evaluate diurnal differences (before-noon vs after-noon) across groups defined by limb dominance and BCRL status (International Society of Lymphology [ISL] stage 0 vs stage 1).

RESULTS: Over 12 months, ECW/TBW ratios measured by the home-based device demonstrated strong agreement with in-clinic measurements, showing minimal bias and high ICC/CCC values. Longitudinal analysis revealed that ECW/TBW changes did not follow uniform patterns within ISL stage categories, showing substantial physiological heterogeneity even among clinically stable groups. Diurnal analysis identified a small but statistically significant decrease in ECW/TBW ratios in the afternoon (P<.001). The magnitude of this decrease differed by limb dominance and BCRL status, with the most pronounced reduction observed in participants whose dominant arm was affected and who had a history of stage 1 lymphedema. ECW/TBW variability was driven more by within-individual factors (eg, measurement timing) than by between-individual differences.

CONCLUSIONS: Home-based segmental bioimpedance provides reliable longitudinal data and reveals granular fluid patterns not captured by conventional ISL staging alone. The significant impact of diurnal variation, particularly in relation to limb dominance, underscores the need for standardizing measurement protocols. Standardizing home-based measurements to a fixed monitoring time can minimize physiological noise and enhance the interpretability of long-term self-monitoring strategies for breast cancer survivors.

PMID:42085680 | DOI:10.2196/86624

JMIR Public Health Surveill. 2026 May 5;12:e89671. doi: 10.2196/89671.

ABSTRACT

BACKGROUND: With global aging, the burden of tuberculosis (TB) among older adults escalates, yet spatial studies on this group are scarce. In Chongqing, where 18.87% of the population are aged 65 years and older and TB burden is high, controlling older adult TB remains a major challenge.

OBJECTIVE: This study analyzed the spatiotemporal patterns of TB among adults aged 65 years and older in Chongqing, China, to inform local prevention and control strategies.

METHODS: The study data were obtained from the Tuberculosis Information Management System of China. Global and local spatial autocorrelation analyses were conducted using ArcGIS (version 10.7) to identify high-risk spatial clusters and visualize their distribution. Spatiotemporal scan statistics were performed using SaTScan (version 10.3.2) to detect clusters of TB cases among the older adult population. Statistical significance was set at P<.05.

RESULTS: The average annual incidence of TB among older adults in Chongqing was 69.59 per 100,000 population, with peaks occurring in spring and summer. The global Moran I ranged from 0.618 to 0.756 (P<.001 in all cases), indicating significant clustering. Persistent high-risk areas were identified in the northeastern and southeastern parts of Chongqing. Spatiotemporal scan statistics detected 1 most likely cluster (relative risk=3.52, 95% CI 3.37-3.68; log-likelihood ratio=1017.43; P<.001) and 3 secondary clusters.

CONCLUSIONS: Significant seasonal patterns of TB among older adults were observed in Chongqing. High-risk areas were predominantly concentrated in the northeastern and southeastern parts of the municipality. More targeted public health interventions are imperative.

PMID:42085675 | DOI:10.2196/89671

JMIR Hum Factors. 2026 May 5;13:e81638. doi: 10.2196/81638.

ABSTRACT

BACKGROUND: More than 20% of perinatal women experience depression, with suicide being a leading cause of maternal death in the United States. Professional societies emphasize the need to identify those at risk of developing perinatal depression to better target preventive care delivery during pregnancy.

OBJECTIVE: We evaluated receptivity to a machine learning-based predictive screener designed to identify women in the first trimester of pregnancy who were asymptomatic but were at risk for developing moderate to severe depression symptoms later in pregnancy.

METHODS: Our participants were adult pregnant women with negative first-trimester depression (Patient Health Questionnaire-9) screens at 1 of 4 obstetric practices. Of the 810 women who were clinically eligible, 787 were successfully contacted via their patient portal. Of these, 289 (36.7%) viewed the screener and 255 (88.2%) completed the 6-question predictive screener. In total, 51 (20%) were identified by the screener as being at risk for developing perinatal depression. Participants were asked a series of follow-up questions regarding the acceptability of the predictive screener and desired preventive resources. Chi-square tests were used to compare demographic characteristics, perceived benefits and concerns, and desired resources between those identified as at risk for depression and those who were not. Differences in acceptability ratings between the two risk groups were determined using nonparametric Mann-Whitney U tests.

RESULTS: On a 5-point Likert scale of agreement, participants found the screener questions easy to complete (median score 5, IQR 5-5) and felt comfortable sharing their answers with their obstetric care providers (median 5, IQR 4-5). Key perceived benefits of completing the screener included opportunities to seek preventive care (75/255, 29.4%) and to receive education on depression risk (66/255, 25.9%). Primary concerns about knowing one’s risk of future depression included worrying about developing depression (90/255, 35.3%) and a lack of prevention opportunities (39/255, 15.3%). Desired preventive resources included counseling (197/255, 77.3%), mind-body interventions (166/255, 65.1%) such as exercise, and prenatal classes or support groups (81/255, 31.8%).

CONCLUSIONS: Participants found the screener acceptable and felt comfortable receiving it through their patient portal. Specific preventive care options were commonly endorsed, several of which are scalable and evidence based. A minority of participants voiced addressable concerns about knowing their risk of developing depression in the future.

PMID:42085674 | DOI:10.2196/81638

JMIR Form Res. 2026 May 5;10:e82667. doi: 10.2196/82667.

ABSTRACT

BACKGROUND: The growing integration of artificial intelligence (AI) in higher education has transformed learning processes but also raised concerns about potential mental health risks. Medical students represent a particularly vulnerable group due to high academic stress and increasing reliance on generative AI tools for study and decision-making tasks. Despite this, the relationship between AI dependence and psychological distress remains underexplored in Latin American contexts.

OBJECTIVE: This study aimed to evaluate the association between generative AI dependence and levels of stress, anxiety, and depression among medical students.

METHODS: A cross-sectional study was conducted with 187 human medicine students from a Peruvian university during the first academic semester of 2025. The Dependence on Artificial Intelligence Scale and the Depression, Anxiety, and Stress Scale-21 were applied. Negative binomial regression models, both crude and adjusted for sex, age, income, and year of study, were used to assess associations, reporting rate ratios (RRs) and 95% CIs.

RESULTS: Participants had a median age of 22 (IQR 19-24) years, and 58.8% (110/187) were female. The median Dependence on Artificial Intelligence Scale score was 10 (IQR 7-14). Generative AI dependence showed significant correlations with anxiety (ρ=0.336, 95% CI 0.22-0.44) and depression (ρ=0.316, 95% CI 0.20-0.43) and a smaller correlation with stress (ρ=0.277, 95% CI 0.16-0.39). In the adjusted regression models, each 1-point increase in generative AI dependence was associated with a 5% higher expected anxiety score (RR 1.05, 95% CI 1.01-1.09; P=.01) and a 4% higher depression score (RR 1.04, 95% CI 1.01-1.08; P=.03), whereas the association with stress was positive but nonsignificant (RR 1.03, 95% CI 1.00-1.07; P=.08). Fifth-year students had significantly greater anxiety levels than their sixth-year peers (RR 1.82, 95% CI 1.09-3.01; P=.02). No significant effects were observed for sex, age, or income.

CONCLUSIONS: This study empirically examined generative AI dependence as a distinct behavioral construct and its association with mental health symptoms in medical students. Unlike prior research, this study evaluated psychological dependence on generative AI and modeled its relationship with anxiety and depression using appropriate count-based regression techniques. By providing early evidence from a Latin American context, it contributes to the emerging field of digital mental health and medical education research. These findings underscore the need for universities to promote balanced and responsible AI use, integrate digital literacy with mental health support strategies, and develop preventive policies that mitigate potential maladaptive reliance on generative AI tools.

PMID:42085672 | DOI:10.2196/82667

JMIR Form Res. 2026 May 5;10:e75133. doi: 10.2196/75133.

ABSTRACT

BACKGROUND: Wearable activity trackers are useful tools to track and monitor physical activity (PA), especially considering their use in free-living environments. Users often see moderate improvements in step count, but consistent increases at various intensities of PA are inconclusive. While wearable research is growing, no known studies specifically examine the relationship between how the use of self-selected social features on wearables affects PA and exercise self-efficacy.

OBJECTIVE: This study aims to compare weekly PA, approximating moderate-to-vigorous intensity, of adults from the New York City metropolitan area assigned to either use or not use social engagement PA features on their device. Exercise self-efficacy was also measured. Additionally, a preliminary examination into the use of 3 different social features was conducted to inform where controlled parameters on feature use may be needed in future work.

METHODS: The researchers conducted a real-world pilot study by recruiting wearable users aged 18 years and older in the New York City area to wear their devices in free-living environments. After consent, participants were randomized into 1 of 2 conditions: the condition that involved use of the social engagement PA features or the condition that did not for 8 weeks. Participants submitted objective data from their device and completed a self-efficacy measure at baseline, week 4, and week 8. Those in the intervention group also answered questions about which social feature they used the most throughout the study.

RESULTS: Data from 123 participants were analyzed using mixed methods analysis. Principal findings included no difference between wearable social feature users and nonusers in weekly PA (P=.55) or exercise self-efficacy (P=.47). There was an overall effect of time across the repeated measures on PA (P=.006) with an average increase of 72 (SD 3) minutes. Secondary findings highlight the need to control for the use of only a single social feature to identify more concrete effects. An effect of time was found across the repeated measures (P=.01) in the intervention group, showing an increase of 49 to 126 minutes of PA, depending on the feature used most. The mixed methods analysis also found that exercise self-efficacy did not significantly change based on which social feature was used most (P=.24).

CONCLUSIONS: Consistent with other literature, this pilot study demonstrates that using wearables can lead to increases in PA and that sharing one’s PA data with others may amplify the effect. However, the novelty of this study is that although carefully implied, specific social features on a wearable may have a greater effect than others. This study identified the need for further investigation into which features may be more effective. With the increased prevalence of device ownership, knowing if certain social features lead to greater increases in PA may help those encouraging PA behavior change.

PMID:42085670 | DOI:10.2196/75133