Tag: statistics

Forensic Sci Int. 2026 Apr 29;386:112980. doi: 10.1016/j.forsciint.2026.112980. Online ahead of print.

ABSTRACT

A high-level data fusion approach for classifying spray paint samples from five major U.S. manufacturers, each represented by five color groups (black, blue, red, silver/gray, and white), was investigated. Spectral data were collected using four analytical techniques: Fourier transform infrared (FTIR) spectroscopy, Raman spectroscopy with laser excitation at 532 nm and 785 nm, scanning electron microscopy coupled with energy-dispersive spectroscopy (SEM-EDS), and UV-Vis microspectrophotometry (MSP). Their combined use and discriminating ability were evaluated. Each dataset was independently modeled using five supervised machine learning classifiers: Naïve Bayes, k-nearest neighbors (KNN), support vector machine (SVM), random forests, and extreme gradient boosting (XGBoost). The intermediate predictions from each classifier were integrated using the majority voting mechanism to yield a final class assignment, forming a high-level data fusion scheme. The proposed approach consistently outperformed individual instruments, achieving near-perfect classification accuracy across several color blocks, particularly for red and blue paints. Among classifiers, generally, Random Forest and Naïve Bayes provided the most stable performance, while SVM with a linear kernel and XGBoost showed lower accuracy. The findings confirm that fusing complementary spectral information improves discriminative ability, reduces redundancy, and creates a computationally efficient, reproducible framework for objective evaluation of source-level questions arising from forensic paint examinations. Overall, the developed model mirrored the process followed by forensic paint examiners in recognizing relevant spectral features from the various techniques. This approach offers a promising pathway toward integrating multimodal spectral data within probabilistic or likelihood ratio-based frameworks following comparative examinations of paint.

PMID:42096743 | DOI:10.1016/j.forsciint.2026.112980

Comput Biol Chem. 2026 May 5;124(Pt 1):109097. doi: 10.1016/j.compbiolchem.2026.109097. Online ahead of print.

ABSTRACT

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) enables rapid species identification in clinical microbiology and shows promise for predicting antimicrobial resistance (AMR) from mass spectra. However, models trained at one hospital site suffer substantial performance degradation at another site due to differences in instruments, sample preparation, and patient populations. Despite numerous domain adaptation (DA) methods in the machine learning literature, none has been systematically benchmarked for cross-site MALDI-TOF AMR prediction. This study presents the first comprehensive benchmark evaluating 13 methods (spanning baselines, supervised transfer learning, and unsupervised DA) across five transfer scenarios involving three public datasets (DRIAMS, MS-UMG, and MARISMa) covering up to 20 species-antibiotic pairs. In total, the benchmark comprises over 15,000 experiments with five random seeds per configuration. A label-efficiency analysis across all five scenarios further examines how model performance scales with 10%, 25%, 50%, and 75% of available target-site labels. The results demonstrate that simple fine-tuning with target-site labels closes 92%-97% of the domain gap and dominates all unsupervised DA methods, which yield only 0%-6% improvement over source-only baselines. The label-efficiency analysis reveals that for competitive transfer methods, as few as 25% of target labels suffice to recover 81%-94% of full supervised performance on cross-site scenarios. These findings provide practical guidelines for clinical deployment: collecting a modest number of labeled samples at the target site is far more effective than applying sophisticated unsupervised adaptation techniques.

PMID:42096742 | DOI:10.1016/j.compbiolchem.2026.109097

Biomed Pharmacother. 2026 May 6;199:119472. doi: 10.1016/j.biopha.2026.119472. Online ahead of print.

ABSTRACT

Statistics indicate that brain metastases occur in nearly 30% of patients with solid tumors, with lung cancer, breast cancer, and melanoma being the three most common primary sources. Brain metastasis is characterized by the co-evolution of tumor cells co-evolve with the brain microenvironment, inducing changes in the phenotype of brain stromal cells that facilitate the colonization, survival, and growth of tumors. The prognosis for brain metastases remains poor, with 2-year and 5-year survival rates for patients diagnosed with brain metastases of 8.1% and 2.4%, respectively. Notably, more than half of brain metastases patients die from neurological diseases. Current treatment options for brain metastases include radiotherapy, neurosurgery, systemic chemotherapy, targeted therapy, and immunotherapy, which are often used in combination to improve therapeutic outcomes. In recent years, the use of nanomaterials for brain metastases treatments has been progressively developed to enhance the efficiency and precision of drug delivery. In the future, with a deeper understanding of the mechanisms underlying brain metastases and further development of treatment strategies, patient outcomes are expected to improve. This review, summarizes the current understanding of brain metastasis mechanisms and therapeutic approaches, and outlines an outlook on future research directions in the field.

PMID:42096737 | DOI:10.1016/j.biopha.2026.119472

Obstet Gynecol. 2026 May 7. doi: 10.1097/AOG.0000000000006292. Online ahead of print.

ABSTRACT

OBJECTIVE: To evaluate the driving distance from U.S. prisons housing women to hospitals providing obstetric care and advanced neonatal intensive care.

METHODS: Using the Google Distance Matrix API, we conducted a cross-sectional analysis to calculate driving distances from state and federal prisons housing women to the closest in-state obstetric hospitals and level III or IV neonatal intensive care units (NICUs). The primary outcome was obstetric hospital distance (driving distance from each prison to the closest hospital providing obstetric care). The secondary outcome was NICU hospital distance (driving distance from each prison to the closest level III or IV NICU). We calculated state-level descriptive statistics and assessed regional differences using the Kruskal-Wallis test.

RESULTS: Of 136 prisons, 134 (98.5%) had an in-state driving route to an obstetric hospital. There were 1,920 obstetric hospitals and 836 level III or IV NICUs. The median (IQR) obstetric hospital distance was 11.4 miles (4.5-22.8) (range 0.6-139.8). The farthest obstetric hospital distances were in Wyoming (139.8 miles), North Dakota (122.4 miles), and Georgia (69.6 miles). Twelve prisons (9.0%) were located more than 37.2 miles from the closest obstetric hospital, including two in both Wyoming and Georgia. Of the 130 prisons with driving routes to an in-state level III or IV NICU, the median (IQR) NICU hospital distance was 19.5 miles (7.7-39.6) (range 0.4-357.9). The farthest NICU hospital distances were in Alaska (357.9 miles), New Mexico (187.0 miles), and South Dakota (175.8 miles). Distances to the closets NICU were significantly longer in the South and Midwest than in the Northeast (P=.025).

CONCLUSION: Although most prisons housing women were located near hospitals providing obstetric and neonatal intensive care, there were state and regional disparities. For some prisons, distance could pose a substantial barrier to timely perinatal care and may exacerbate existing inequities in maternal and neonatal morbidity and mortality.

PMID:42096711 | DOI:10.1097/AOG.0000000000006292

N Z Med J. 2026 May 8;139(1634):51-64. doi: 10.26635/6965.7460.

ABSTRACT

BACKGROUND: Gastric cancer (stomach cancer) is an important contributor to morbidity and mortality in Aotearoa New Zealand, with marked ethnic inequities. Although national incidence rates are declining, Māori and Pacific peoples continue to experience higher rates than other groups. Demographic change and regional population growth are expected to influence future burden, yet no published projections provide estimates disaggregated by ethnicity and region.

METHODS: Gastric cancer registrations from 2001 to 2022 from the New Zealand Cancer Registry were linked to population estimates and projections stratified by age, sex, prioritised ethnicity and Health New Zealand – Te Whatu Ora region. Incidence was modelled using an age-period-cohort approach with time-based weighting to emphasise recent trends. Projections to 2045 were generated, and uncertainty was quantified using 1,000 non-parametric bootstrap iterations incorporating perturbation of population denominators.

RESULTS: Gastric cancer cases are projected to increase by 47.7% to approximately 725 per year by 2045, despite a decline in the age-standardised rate from 5.9 to 5.3 per 100,000. All regions show increasing absolute numbers, with the Northern Region experiencing the largest rise. Māori and Pacific peoples have the highest current incidence and a large proportional increase in projected cases, although incidence rates decline modestly for all ethnic groups. Future case growth is driven mainly by demographic expansion and an ageing population.

CONCLUSION: Absolute gastric cancer cases are projected to increase, particularly among Māori and Pacific populations and in regions experiencing rapid population growth. This has implications for early diagnosis and specialist service delivery. These projections support equity-focussed prevention and service planning, including Helicobacter pylori control, timely diagnostic pathways, and regional planning for specialist cancer services.

PMID:42096700 | DOI:10.26635/6965.7460

N Z Med J. 2026 May 8;139(1634):32-37. doi: 10.26635/6965.7068.

ABSTRACT

AIM: This study aims to describe the epidemiology, clinical features, microbiology and management of paediatric patients (<18 years) admitted to Christchurch Hospital with periorbital or orbital infections over a 10-year period.

METHODS: A retrospective review was conducted of all patients under 18 years admitted with periorbital and orbital infections between 2013 and 2023. Cases were identified using surgical theatre records and discharge coding, with data extracted from electronic medical records. Clinical, demographic, microbiological and management data were analysed descriptively.

RESULTS: A total of 495 paediatric cases were identified, with 93% presenting with periorbital cellulitis and 7% with orbital cellulitis. Sinusitis was the predominant predisposing factor for postseptal disease, present in 83% of those cases. Orbital signs such as proptosis, pain with eye movement, reduced visual acuity and ophthalmoplegia were more frequent in orbital cellulitis. Orbital cases had longer hospital stays with a median of 4.5 days (range 2-33 days) compared to periorbital disease with a median of 1 day (range 0-8 days). Orbital cases also had a higher rate of surgical intervention (47%), most commonly functional endoscopic sinus surgery. Staphylococcus aureus was the most frequently isolated organism in both groups (45% periorbital, 42% orbital). Māori and Pacific children were disproportionately affected (comprising 20% and 10% respectively of periorbital cases and 17% and 19% of orbital cases).

CONCLUSIONS: The presence of orbital signs should prompt urgent imaging to exclude orbital disease. Sinusitis remains a key risk factor for orbital cellulitis, and Māori and Pacific children are disproportionately affected.

PMID:42096698 | DOI:10.26635/6965.7068

N Z Med J. 2026 May 8;139(1634):24-31. doi: 10.26635/6965.7279.

ABSTRACT

AIM: The aim of this article was to assess agreement between verified self-reported fractures in a clinical trial with Accident Compensation Corporation (ACC) claim data.

METHODS: In a 10-year randomised controlled trial of 1,054 women aged 50-60 years, participants self-reported fractures as they occurred or on routine 6-monthly questionnaires. Radiology imaging and reports were used to verify fractures, which were then compared with ACC claims data (ACC is the New Zealand no-fault accident claims organisation funded through levies). Initially, fracture claim data only were obtained, followed by all ACC claims for each participant for the study period.

RESULTS: Three hundred and fifty-six self-reported fractures in 248 women were verified in the trial, whereas there were 328 ACC fracture claims from 238 women for the study period. Out of 356 trial fractures, 211 (59%) had a matching ACC fracture claim, and out of 328 ACC fracture claims 211 (64%) had a matching trial fracture. After obtaining all ACC claims, we identified a matching ACC claim for 340/356 (96%) trial fractures: 59% were fracture claims and 31% soft-tissue injury claims.

CONCLUSIONS: Repurposing ACC fracture claims data for clinical trials has significant limitations and is likely to introduce false negative and false positive events. When tolerance for misclassification is higher (e.g., large non-randomised studies), ACC claims data may be useful because 60% of claims had a verified fracture, with higher proportions for major fracture types.

PMID:42096697 | DOI:10.26635/6965.7279

N Z Med J. 2026 May 8;139(1634):12-23. doi: 10.26635/6965.7169.

ABSTRACT

AIM: To understand the impacts and responses of households during the Havelock North drinking water outbreak.

METHODS: Fifty days after the outbreak, cross-sectional telephone questionnaires were administered to a cohort of households.

RESULTS: Seventy-six percent of the people surveyed indicated drinking unboiled tap water, with 35% of those developing diarrhoea, compared with only 3% of those who did not drink the water. Symptoms correlated with increasing quantities of water consumed, and 31% reported a relapse of diarrhoea after initial improvement. The attack rate among those less than 20 years old (41%), was higher than those aged 50 and over (22%). Individuals with diarrhoea had an average of 7 days off school or work. Only 27% of individuals with diarrhoea visited a doctor or hospital, but 72% were in households that purchased items from a pharmacy. Following the issue of a boil water notice, 82% of households boiled their water, and 67% purchased bottled water, with only 5% taking no precautions. A third of the 169 households surveyed continued one or both of these responses for at least 3 weeks after the boil water notice was lifted.

CONCLUSIONS: Telephone surveys provided insights into the outbreak not otherwise obtainable from routine surveillance systems, including the attack rates among different demographics, size of the outbreak (5,540 cases within Havelock North), potential of pharmacy-based surveillance, compliance with public health messaging and the need to communicate to households when the water is safe to drink.

PMID:42096696 | DOI:10.26635/6965.7169

J Med Internet Res. 2026 May 7;28:e85098. doi: 10.2196/85098.

ABSTRACT

BACKGROUND: Military sexual trauma (MST) can have significant adverse effects on mental health and well-being, often leading to posttraumatic stress disorder (PTSD) symptoms and maladaptive beliefs. Although effective psychotherapies exist, stigma, confidentiality concerns, and systemic barriers often hinder help-seeking among service members and veterans. Mobile mental health apps offer an accessible and anonymous support alternative, potentially addressing such barriers. However, app effectiveness depends on user engagement and emerging evidence suggests that engagement may be shaped by symptom severity.

OBJECTIVE: This retrospective observational study aimed to explore the relationship between posttraumatic symptom severity and user engagement with Beyond MST (US Department of Veterans Affairs [VA] National Center for PTSD), an app for individuals who experienced MST. Specific aims included (1) characterizing trauma-related symptom levels and app engagement among users who completed in-app assessments, and (2) evaluating how PTSD symptom severity, negative posttraumatic cognitions, and mental well-being relate to objective measures of engagement.

METHODS: Anonymous usage data from 27,517 users collected between March 11, 2021 and July 29, 2024, were analyzed. Three subsamples were identified: those who completed the in-app PTSD checklist for DSM-5 (Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition]; PCL-5, n=3689), the Posttraumatic Maladaptive Beliefs Scale (PMBS; n=2197), and the Warwick-Edinburgh Mental Well-Being Scale (WEMWBS; n=2160). Engagement metrics included duration of use (ie, days of use and minutes of use), frequency of feature access (ie, coping tool and psychoeducation access), and frequency of feature use (ie, total assessment completions). Regression analyses, including quadratic terms, were conducted to evaluate how symptom severity and well-being levels influenced engagement and identify possible curvilinear trends.

RESULTS: Median engagement levels ranged across subsamples as follows: 3-4 days of use (IQR 5-6), 22-30 minutes of use (IQR 33.7-42.9), 1-5 feature accesses (IQR 6-9), and 2-3 assessment completions (IQR 2). Subsamples were highly symptomatic. Analyses revealed that moderate PTSD symptom and negative posttraumatic cognition severity were associated with higher engagement relative to users with very low and very high symptom levels, particularly for days of use and frequency of coping tool access. Conversely, higher mental well-being scores were generally linked to increased app engagement with linear effects. Effect sizes were small, suggesting limited clinical impact.

CONCLUSIONS: This study highlights the possible challenges in engaging highly symptomatic individuals with digital mental health interventions. Although Beyond MST successfully reaches its targeted population, very low or high symptom levels and lower well-being may hinder sustained engagement. These findings suggest that symptom levels should be considered in app development (ie, personalization) and when integrating apps into professional care. Interpretation is limited by the anonymous nature of the data, which prevented characterization of users and their trauma histories. Further research is needed to clarify how symptom patterns influence engagement, especially in trauma contexts.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.31979/etd.882a-5fcx.

PMID:42096694 | DOI:10.2196/85098

JMIR Med Inform. 2026 May 7;14:e78030. doi: 10.2196/78030.

ABSTRACT

BACKGROUND: The expansion of telehealth services, particularly during the COVID-19 pandemic, has transformed health care delivery in the United States. Telehealth promises greater access and resource efficiency by reducing wait times and appointment lengths, especially in specialties like psychiatry, behavioral health, bariatrics, and sleep medicine. However, disparities exist in adoption based on demographics, geography, and socioeconomic status, raising concerns about equitable access and optimal resource use.

OBJECTIVE: This study aims to evaluate how telehealth impacts health care resource use across 4 specialties by examining 2 key metrics: patient-to-provider ratios and appointment durations. It seeks to understand how factors such as patient demographics, facility characteristics, and social determinants influence telehealth adoption and efficiency using a national dataset spanning from 2018 to 2023.

METHODS: We analyzed a deidentified dataset from Epic Cosmos, covering outpatient visits across 48 US states (2018-2023). After data preprocessing and feature engineering, we applied 3 machine learning (ML) models (random forest, extreme gradient boosting, and deep neural networks) to predict resource use. Using the model performing the best, feature importance was assessed using Shapley Additive Explanations values. We then used k-means clustering to group facilities into clusters per specialty. Comparative analyses were conducted to evaluate differences in use among clusters, during and after the pandemic.

RESULTS: Telehealth use peaked in 2020 and has remained above prepandemic levels since then. In 2018-2023, telehealth adoption reached 36.9% (4,543,021/12,311,710) in psychiatry, 23.9% (5,321,099/22,264,013) in behavioral health, 21.2% (924,333/4,360,061) in bariatrics, and 16.8% (851,803/5,070,256) in sleep medicine. Telehealth visits were consistently shorter than office visits (mean reduction 12.24 minutes; SD 3.33 minutes; P=.18), while patient-to-provider ratios varied significantly across specialties. Among ML models, extreme gradient boosting regression achieved the best performance (patient-to-provider ratios: R²=0.96-0.99; appointment durations: R²=0.61-0.69). Shapley Additive Explanations analysis identified visit type, telehealth use, facility size, rurality, and Social Vulnerability Index household vulnerability as the strongest predictors. Comparative analyses showed significant differences across clusters (all P<.05).

CONCLUSIONS: Telehealth has become a sustainable component of health care, enhancing access and efficiency across both rural and urban areas. However, its impact varies across specialties and regions, highlighting the need for targeted strategies such as staffing support for vulnerable populations, infrastructure investments in rural facilities, and reimbursement models that reflect telehealth’s resource use. This study provides robust evidence from ML and clustering analyses, demonstrating how telehealth shapes resource use and offering actionable insights for equitable and sustainable integration.

PMID:42096693 | DOI:10.2196/78030