Category: Statistics

Ann Biomed Eng. 2025 Nov 17. doi: 10.1007/s10439-025-03890-0. Online ahead of print.

ABSTRACT

PURPOSE: Digital twin (DT) cohorts are collections of models where each member represents an individual real-world asset. DT cohorts can be used for in-silico trials, outlier detection and forecasting, and are used across engineering, industry, and increasingly in personalised medicine. To increase the scalability of DT cohorts, researchers often train emulators to be used as cheap surrogates of computationally expensive mathematical models. Frequently, each cohort member is emulated individually, without reference to other members. We propose that instead, we can treat each DT as a thread in a larger network, and that these threads can be woven together into a digital tapestry using cohort learning methods.

METHODS: We propose two statistical approaches for transferring knowledge between threads. The first method, ‘latent-feature emulators’, utilises a latent representation of individual cohort members to generate a single emulator for the entire cohort. The second method, ‘discrepancy emulators’, learns the discrepancy between a new cohort member and existing members.

RESULTS: In two cardiac DT case studies, we show that these methods can reduce computational costs by more than 50% compared to the standard approach of training individual emulators, even in small cohorts.

CONCLUSIONS: We find that by transferring information between meshes, the cohort methods improve both the computational efficiency and the accuracy of emulators when compared to the standard approach of individually emulating each cohort member. As cohort size increases, the computational savings grow further. We focus on the use of Gaussian process emulators, but the transfer methods are applicable to other surrogate approaches such as neural networks.

PMID:41249625 | DOI:10.1007/s10439-025-03890-0

Cancer Gene Ther. 2025 Nov 17. doi: 10.1038/s41417-025-00988-4. Online ahead of print.

ABSTRACT

Thyroid cancer (TC) is an endocrine malignancy characterized by metabolic abnormalities, with its incidence continually on the rise. Understanding the pathogenesis of this cancer would help develop better diagnostic and therapeutic methods. We aimed to integrate single-cell transcriptomics, bulk transcriptomics, and GWAS data to identify causal associations with thyroid cancer at the gene level. We intended to utilize single-cell atlases to identify malignant cells and their characteristics, and employed SMR to search for genetic loci causally associated with thyroid cancer. We validated the expression differences of the genes at the single-cell level and bulk level, as well as through immunohistochemistry experimental results. We investigated the tumor immune microenvironment of patients, attempting to find immune subgroups with differential proportions. Based on these subgroups, we conducted multi-machine learning modeling to predict the likelihood of disease and developed a corresponding interactive web application. HMGA2, SDCCAG8, DLG5, MT1E, RABL2B, RERE, and NDUFA12 all demonstrated to varying degrees their roles in promoting or inhibiting the occurrence and development of thyroid cancer, with HMGA2 showing consistency across all analyses. We also identified some immune subtypes significantly associated with TC and chose markers of T_cell_C8_STMN1 to construct patient diagnostic models. Through various combinations of machine learning feature selection and model construction, we ultimately built 178 diagnostic models, with the combination of glmBoost+RF having the best diagnostic performance (Average AUPR: 0.9915). The predictive web pages ( https://zclab-cnp.shinyapps.io/TC-WEB/ ) can provide convenience and reference for clinical personnel.

PMID:41249621 | DOI:10.1038/s41417-025-00988-4

J Clin Immunol. 2025 Nov 18;45(1):162. doi: 10.1007/s10875-025-01959-y.

ABSTRACT

Flow cytometric immunophenotyping of lymphocytes and dendritic cells, and functional lymphocyte mitogen response tests are used in the diagnostics of inborn errors of immunity (IEI), especially in pediatrics. These routinely used tests lack sufficient age-matched reference values in children. We established reference values for lymphocyte and dendritic cell subsets for four age groups from 68 healthy children under 12 years of age. These values were then compared to prior publicly available articles and 46 clinical samples from children with confirmed IEI diagnosis. Mitogen response results were also compared between 27 children and 177 adults. In the literature review, we found considerable variability in lymphocyte subset definitions and statistical approaches. Most IEI patients had increased transitional and naïve B, and decreased memory B cells. CHH patients had increased γδ T and DNTs. Lymphocyte stimulation via FASCIA method provides weaker stimulation results in children than in adults, which seems to result from a larger proportional count of naïve lymphocytes in children. The established reference values can be used in diagnostics of pediatric immunological conditions in laboratories that use similar analytic methods. Lower lymphocyte mitogen response results in children need to be taken into consideration when interpreting the results of lymphocyte functional tests.

PMID:41249610 | DOI:10.1007/s10875-025-01959-y

Discov Oncol. 2025 Nov 17;16(1):2114. doi: 10.1007/s12672-025-03886-1.

ABSTRACT

BACKGROUND: Uterine leiomyosarcoma is a rare and aggressive malignancy with limited responsiveness to standard therapies. We conducted a real-world, multicenter study to compare the clinical efficacy and safety of two commonly used first-line chemotherapy regimens-doxorubicin-ifosfamide and gemcitabine-docetaxel-in patients with metastatic uterine leiomyosarcoma.

METHODS: This retrospective cohort included 271 patients with advanced or metastatic uterine leiomyosarcoma treated between 2010 and 2023 across 30 centers in Turkey. Patients received either doxorubicin-ifosfamide (n = 142) or gemcitabine-docetaxel (n = 129) as first-line therapy. The primary endpoint was overall survival; secondary endpoints included progression-free survival, objective response rate, disease control rate, and safety. Adverse events were graded according to the Common Terminology Criteria for Adverse Events version 5.0, while survival outcomes were estimated using the Kaplan-Meier method and further analyzed with Cox proportional hazards models.

RESULTS: Median overall survival was 19.7 months with doxorubicin-ifosfamide and 20.2 months with gemcitabine-docetaxel (P = .26). Median progression-free survival was 5.5 months with doxorubicin-ifosfamide and 7.0 months with gemcitabine-docetaxel (P = .62). The objective response rate was numerically higher with gemcitabine-docetaxel (35% vs. 26%), although not statistically significant (P = .11). Grade 3-4 neutropenia (16% vs. 12%) and febrile neutropenia (7% vs. 6%) were more frequent with doxorubicin-ifosfamide.

CONCLUSIONS: In this largest-to-date real-world cohort of metastatic uterine leiomyosarcoma, doxorubicin-ifosfamide and gemcitabine-docetaxel demonstrated comparable survival outcomes. Gemcitabine-docetaxel, however, was associated with a more favorable hematologic safety profile. These findings support the clinical utility of both regimens while underscoring the need for prospective, biomarker-driven studies to refine treatment selection and improve personalization in this rare malignancy.

PMID:41249604 | DOI:10.1007/s12672-025-03886-1

NPJ Precis Oncol. 2025 Nov 17;9(1):355. doi: 10.1038/s41698-025-01146-7.

ABSTRACT

The treatment of Epithelial Ovarian cancer (EOC) could benefit from the addition of bevacizumab (BEV) to standard chemotherapy in selected patients. Gene expression (GE) profiling and the evaluation of immune infiltration are used to define patients’ prognosis. However, their role as prognostic and/or predictive biomarkers for the efficacy of antiangiogenic therapy efficacy remains uncertain. In this study, we combined GE profiling and multiplex immunofluorescence (MIF) analyses on material from patients enrolled in the phase IV MITO16A/MaNGO OV-2 trial, assessing associations between immune infiltrate and patients’ prognosis. Patients were stratified into four molecular subtypes, and CIBERSORTx was applied to infer the cell-type-specific expression pattern of immune populations. MIF evaluated the presence of immune cells in the tumor and stromal compartments. These complementary experimental approaches revealed that immune infiltration is associated with shorter progression-free survival in BEV-treated patients, warranting future investigation to evaluate its use as a viable biomarker for patient stratification. Trial registration: NCT01706120, EudraCT number: 2012-003043-29, Date of registration 24 September 2012.

PMID:41249601 | DOI:10.1038/s41698-025-01146-7

J Perinatol. 2025 Nov 17. doi: 10.1038/s41372-025-02485-w. Online ahead of print.

ABSTRACT

OBJECTIVE: To evaluate the effectiveness of implementing American Academy of Pediatrics (AAP) discharge guidelines in reducing unplanned hospital readmissions within 30 days post-discharge among term ansd late preterm newborns.

STUDY DESIGN: Retrospective observational study analyzing unplanned readmissions at a single-center neonatal unit from January 1, 2021, to December 31, 2024. Data were compared before (January 1, 2021-June 30, 2022) and after (July 1, 2022-December 31, 2024) guideline implementation, with subgroup analysis for the period after addition of structured support (July 1, 2023-December 31, 2024).

RESULT: AAP guideline implementation was associated with a statistically significant reduction in unplanned readmission rates among term infants (0.66% vs. 0.33%; P = 0.008). No reduction was evident among late preterm infants. Subgroup analysis showed further reductions post-structured support addition, though confounding by provider changes limits attribution.

CONCLUSION: The adoption of the AAP discharge guidelines, along with a structured process of mother and infant readiness, significantly decreassed unplanned readmission rates among term newborns.

PMID:41249592 | DOI:10.1038/s41372-025-02485-w

Biomed Phys Eng Express. 2025 Nov 17. doi: 10.1088/2057-1976/ae202c. Online ahead of print.

ABSTRACT

Sleep arousal, characterized by emergence of light sleep or partial wakefulness, often indicates underlying physical disorders, and its detection is crucial for effective patient treatment. While the detection of arousals using multiple signals can be effective, the dependencies on multiple electrodes impose burden on patients. To resolve this issue, some effective features estimated from single-lead electroencephalography (EEG) signals were proposed to detect sleep arousal. Normalized and filtered EEG signals were segmented into 7-second frames, and scalograms were estimated using continuous wavelet transform (CWT). Scalograms and local properties such as frequency, bandwidth, band energy, band energy ratio, maxima, and regularity were derived from the coefficients of CWT. Final classification features were generated using statistical analyses. The most effective features, estimated by correlation coefficients and p-values, were subjected to an artificial neural network to evaluate the performance of the features. The maximum classification performances (86.72% accuracy, 89.26% sensitivity, 86.55% specificity, and 94.87% AUC) were achieved with 100 features. However, sixty specific features were selected from a total of 182 classification features, yielding nearly the same performance as the maximum. Finally, only 14 features were identified as making a pronounced contribution to arousal detection. These findings highlighted the potential of a feature-efficient single-channel EEG-based approach for reliable sleep arousal detection. The proposed framework can be integrated into patient monitoring systems, such as apnea detection modules, to provide a more comprehensive tool for sleep disorder management.

PMID:41248549 | DOI:10.1088/2057-1976/ae202c

J Educ Eval Health Prof. 2025;22:36. doi: 10.3352/jeehp.2025.22.36. Epub 2025 Nov 18.

ABSTRACT

PURPOSE: This study systematically evaluates and compares the performance of large language models (LLMs) in answering medical licensing examination questions. By conducting subgroup analyses based on language, question format, and model type, this meta-analysis aims to provide a comprehensive overview of LLM capabilities in medical education and clinical decision-making.

METHODS: This systematic review, registered in PROSPERO and following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, searched MEDLINE (PubMed), Scopus, and Web of Science for relevant articles published up to February 1, 2025. The search strategy included Medical Subject Headings (MeSH) terms and keywords related to (“ChatGPT” OR “GPT” OR “LLM variants”) AND (“medical licensing exam*” OR “medical exam*” OR “medical education” OR “radiology exam*”). Eligible studies evaluated LLM accuracy on medical licensing examination questions. Pooled accuracy was estimated using a random-effects model, with subgroup analyses by LLM type, language, and question format. Publication bias was assessed using Egger’s regression test.

RESULTS: This systematic review identified 2,404 studies. After removing duplicates and excluding irrelevant articles through title and abstract screening, 36 studies were included after full-text review. The pooled accuracy was 72% (95% confidence interval, 70.0% to 75.0%) with high heterogeneity (I2=99%, P<0.001). Among LLMs, GPT-4 achieved the highest accuracy (81%), followed by Bing (79%), Claude (74%), Gemini/Bard (70%), and GPT-3.5 (60%) (P=0.001). Performance differences across languages (range, 62% in Polish to 77% in German) were not statistically significant (P=0.170).

CONCLUSION: LLMs, particularly GPT-4, can match or exceed medical students’ examination performance and may serve as supportive educational tools. However, due to variability and the risk of errors, they should be used cautiously as complements rather than replacements for traditional learning methods.

PMID:41248547 | DOI:10.3352/jeehp.2025.22.36

J Public Health Manag Pract. 2026 Jan-Feb 01;32(1S Suppl 1):S122-S129. doi: 10.1097/PHH.0000000000002258. Epub 2025 Nov 18.

ABSTRACT

INTRODUCTION: Conceptualization of the COVID-19 response burden on the state and local government public health workforce is complicated as it continuously evolved and called upon nearly every aspect of the workforce. This study aims to catalog the scale of COVID-19 response among the state and local public health workforce by assessing the role of overtime in the delivery of public health services during pandemic response and shifts in agency size and program area distribution since the height of the pandemic.

METHODS: This study uses detailed state and local workforce survey data from the Public Health Workforce Interests and Needs Survey 2021 and 2024 administrations. The analytic sample includes the 72.2% of employees that served in a COVID-19 response role at any time during the pandemic (n = 30 914; N = 136 591) and 214 state and local agencies that participated in both years (n = 65 144 unduplicated responses).

RESULTS: In total, overtime equated for the equivalent of 25 000 FTE COVID-19 response employees, one-quarter (25%) of the total FTE COVID-19 response workforce. Of the 214 agencies that participated in both the Public Health Workforce Interests and Needs Survey 2021 and 2024, 41% of agencies (88) saw decreases in staff size overall between those 2 points in time. Shifts in primary program area between 2021 and 2024 are largely driven by changes in the proportion of non-full-time permanent employees.

CONCLUSION: This analysis magnifies the strain on the existing capacity of the public health workforce during the COVID-19 pandemic. These challenges stem from a chronically underfunded and understaffed workforce that was not prepared for surge capacity beyond existing employees. Given that many state and local public health agencies are smaller post-pandemic, it is reasonable to conclude that without large infrastructural changes, the workforce would likely face the same challenges if another pandemic-like crisis were to occur.

PMID:41248538 | DOI:10.1097/PHH.0000000000002258

J Public Health Manag Pract. 2026 Jan-Feb 01;32(1S Suppl 1):S60-S67. doi: 10.1097/PHH.0000000000002233. Epub 2025 Nov 18.

ABSTRACT

OBJECTIVE: Describe key characteristics of the rural local public health workforce on a national level, including in comparison to both the overall and urban local public health workforce.

DESIGN: Cross-sectional analysis of the 2024 Public Health Workforce Interests and Needs Survey (PH WINS) data.

SETTING: Local health departments (LHDs) serving rural and urban jurisdictions across the United States.

PARTICIPANTS: The study sample included 172 679 weighted responses from individuals working in LHDs, and 33 214 of them were from rural-serving LHDs.

MAIN OUTCOME MEASURES: Descriptive and bivariate statistics for measures across 4 areas, both overall and by rurality: demographic characteristics, educational background, position information, and intentions to stay or leave.

RESULTS: Greater portions of the rural local public health workforce were female and White relative to their urban counterparts. Compared to the urban workforce, the portions of the rural workforce without a public health degree and with clinical training were both greater. Tenure in position, agency, and public health practice also differed by rurality, with 19.6% of the rural workforce reporting the greatest tenure in public health practice (21 years or above) compared to 17.8% of the urban workforce. Intentions to stay, leave, or retire also differed by rurality, with 15.4% of the rural workforce reporting intentions to leave in the next year for reasons outside of retirement, compared to 21.6% of the urban workforce.

CONCLUSIONS: Characteristics of the local public health workforce vary by rurality, extending prior research demonstrating differences between rural- and urban-serving LHDs across the nation. Findings should guide rural-focused strategies aimed at strengthening and sustaining the public health workforce.

PMID:41248531 | DOI:10.1097/PHH.0000000000002233