European Journal of Cancer

Background: Cases requiring 13 or more tissue sections in Mohs micrographic surgery (MMS) demand extended operative time, additional resources, and often specialised closure techniques. Pre-operative identification of such cases would improve surgical scheduling, resource allocation, and patient counselling. We aimed to develop and validate a machine learning prediction tool using pre-operative clinical features to identify cases likely to require13 sections. Objectives: To develop and validate machine learning models for predicting which Mohs procedures will require 13 sections, using pre-operative clinical features, and to identify key predictive factors. Methods: We analysed 408 consecutive Mohs procedures with 16 pre-operative clinical variables. Thirty machine learning algorithms were evaluated, including ensemble methods (Stacking, Voting), gradient boosting (XGBoost, LightGBM, CatBoost), neural networks (3-7 layers), support vector machines, and traditional classifiers. Model performance was assessed using 5-fold stratified cross-validation and independent test set evaluation. Feature importance was determined using SHAP (SHapley Additive exPlanations) analysis. Results: The stacking ensemble achieved the highest cross-validation AUC of 0.891 (95% CI: 0.849-0.934) and test AUC of 0.884. Tumour area (cm2), calculated using the ellipse formula to approximate clinical tumour morphology, emerged as the strongest predictor (SHAP importance: 0.141), followed by tumour size dimensions (0.086 and 0.068), aggressive histopathology (0.046), and recurrence status (0.035). Wide neural network architectures (5-layer) outperformed deeper configurations (7-layer). The model demonstrated 70.7% high-confidence predictions with uncertainty <15%. Conclusions: Machine learning models using pre-operative clinical features can accurately predict which Mohs procedures will require 13 or more sections. The stacking ensemble approach provides robust predictions suitable for clinical decision support. External validation in multi-centre cohorts with diverse patient populations and practice patterns is warranted to assess model generalisability.

Background The clinical significance of clonal hematopoiesis of indeterminate potential (CHIP) in melanoma remains incompletely defined, particularly with respect to CHIP genotype, clone size, and somatic mutations (e.g BRAF mutations). We integrated human cohort data and a syngeneic melanoma mouse model to evaluate whether CHIP is associated with melanoma risk, tumor growth, and differential clinical outcomes. Methods We analyzed CHIP prevalence and survival in a large treatment-unselected melanoma cohort (n=2,480), evaluated tumor growth in a syngeneic BRAF-mutant (BRAFmut) melanoma murine model of TET2-CHIP and DNMT3A-CHIP, and assessed survival outcomes in an immune checkpoint inhibitor (ICI)-treated advanced melanoma cohort (n=361). Associations with progression-free survival (PFS) and overall survival (OS) were evaluated using Kaplan-Meier analyses and multivariable Cox proportional hazards models. Results CHIP was enriched among patients with treatment-unselected melanoma compared with age/sex-matched healthy controls, and larger CHIP clone size showed an age-adjusted association with inferior OS. In a syngeneic BRAFmut melanoma murine model, TET2-CHIP, but not DNMT3A-CHIP, was associated with significantly increased primary melanoma tumor growth. Among patients with ICI-treated advanced melanoma, CHIP was associated with worse OS compared with patients without CHIP. TET2-CHIP had the strongest adverse association with survival, whereas DNMT3A-CHIP was not significantly associated with PFS or OS. Conclusions CHIP is enriched in melanoma and exploratory analyses demonstrate genotype-specific differences in melanoma tumor growth and clinical outcomes. These findings support further investigation of genotype-specific CHIP profiling as a potential biomarker for melanoma risk stratification and immunotherapy outcomes.

Population-based mammography screening improves breast cancer outcomes, but its impact on real-world treatment pathways and quality indicators (QIs) remains incompletely described. We conducted a retrospective nationwide cohort study using linked data from the Czech National Cancer Registry and the National Registry of Reimbursed Health Services. Women aged [&ge;]18 years with a first breast cancer diagnosis between 2017 and 2024 were classified as screen-detected (SCR) or diagnostically-detected (DIG) according to the imaging modality preceding histological verification. Outcomes included stage distribution, untreated cases, first-line treatment, main treatment modality, time to treatment, multidisciplinary team discussion (MDT), centralization to Comprehensive Cancer Centres (COCs), and survival patterns. The verified cohort included 47,648 women: 26,817 SCR cases (56.3 %) and 20,831 DIG cases (43.7 %). In this nationwide analysis, SCR breast cancer was associated with earlier stage at diagnosis and better survival patterns, but also with longer time to treatment and longer time to MDT discussion than DIG-detected disease. Although treatment rates were high and centralization improved over time, substantial regional variation persisted in care pathways, MDT use, and access to COCs. These findings support continued strengthening of screening participation, monitoring of care intervals, and quality assurance of MDT reporting and regional oncology care delivery.

The therapeutic landscape for hepatocellular carcinoma (HCC) is evolving rapidly, necessitating scalable approaches to synthesize the expanding scientific literature. We characterized thematic shifts in HCC treatment and prognosis research by conducting a retrospective bibliometric analysis of influential publications from 2023 and 2024. Using the OpenAlex database, we identified the 50 most highly cited papers from each year based on eighteen-month post-publication citation counts. Large language models were deployed to extract, normalize, and classify concepts from unstructured text into canonical topics and parent themes, enabling quantitative year-over-year frequency comparisons. Analysis of these 100 papers revealed a distinct maturation in research focus. Although broad categories like general immunotherapy remained prevalent, their relative frequency declined in favor of specific dual immune checkpoint regimens, notably CTLA-4 inhibition and the durvalumab plus tremelimumab combination. Concurrently, parent themes related to radiomics, imaging, and health systems exhibited significant growth in the 2024 cohort. These findings demonstrate a thematic transition in high-impact HCC research from foundational immuno-oncology toward optimized combination therapies and precision diagnostics. Furthermore, this study highlights the utility of artificial intelligence-driven bibliometrics for objectively tracking dynamic conceptual shifts in oncology. A web interface for exploring the data is available at https://pri.pepkio.com/.

Cutaneous squamous cell carcinoma (cSCC) is the second most common form of cancer worldwide. While most cSCCs are not life-threatening, 2-5% of patients develop metastases. To better understand what causes some cSCCs to progress to metastatic disease, we assembled a nationwide cohort of 19,120 patients with clinico-pathologically annotated tumors linked to metastatic outcome. RNA-sequencing was performed on 378 tumors, and whole-exome sequencing on 147, with balanced numbers of tumors that progressed to metastatic disease (cases) and did not (controls). UV radiation was the dominant mutational signature with additional contributions from aging, APOBEC activity, and, in immunosuppressed patients, azathioprine exposure. We identified 38 genes under selection across a core set of signaling pathways. Gene expression clusters were primarily associated with the differentiation state of tumor cells and secondarily with the composition of the tumor microenvironment. Several mutational and transcriptional programs were associated with metastasis, including a dedifferentiated gene expression signature, activating mutations in the RAS signaling pathway, loss-of-function alterations in the SWI/SNF chromatin remodeling complex, and specific arm-level copy number alterations. A 23-gene expression signature was built to predict metastasis from primary cSCC tissue. The signature was validated in two independent cohorts (N=102 and 52), where it predicted metastasis independently of staging systems. Together, these findings provide the most detailed molecular portrait of cSCC to date and establish an assay for risk stratification suitable for clinical implementation.

Introduction: This study aimed to provide population level survival trends during the era in which new systemic therapies transformed treatment guidelines for metastatic cancer, as well as insights on the real world use of these treatments and associated survival. Methods: Adults diagnosed with synchronous metastatic solid cancer in 2008 until 2022 (22 cancer types) were identified from the Netherlands Cancer Registry. Median overall survival (OS) was assessed by five year diagnostic period. For 2018 until 2022, systemic therapy use in any treatment line was analyzed and survival percentiles within treatment and cancer types were estimated with Kaplan Meier survival analyses. Results: Median OS in the overall cohort (n=280,419 patients) improved from 6 to 8 months between the period 2008 until 2012 and 2018 until 2022. Among patients diagnosed in 2018 until 2022, 15% received immunotherapy, 15% targeted therapy, 29% chemotherapy and/or traditional hormone therapy only, and 39% no systemic therapy. In some cancer types, a relatively large proportion of treated patients had longterm survival (e.g., immunotherapy in melanoma: p50 = 67 months). Other cancer types had a smaller subset of treated patients (p10 and p25) with substantially better outcomes than the median (e.g., targeted therapy in NSCLC: p50 = 22 months, p10 = 96 months). Conclusion: Population level survival for patients with synchronous metastatic solid cancer has modestly improved over time. The marked survival heterogeneity within cancer and treatment types highlights both the potential and uncertainty associated with (novel) treatments. Improved prediction of treatment effects and clear communication regarding survival expectation remain critical. Presenting multiple survival scenarios over median survival alone can support decision making.

Rising costs, slow accrual and molecular substratification of cancers necessitate novel clinical trial designs. We demonstrate that artificial intelligence-generated synthetic patients can replace real controls to reproduce results of the SORAML trial. Using external multimodal data from 1,377 acute myeloid leukemia (AML) patients from previous trials and a real-world registry, we fine-tuned a tabular foundation model to generate synthetic patients, reproducing clinical and genetic features and outcome associations. Synthetic patients were then matched to the original SORAML intervention group using Cox risk scores, replacing the original control and reproducing the original trial result with near-identical median event-free survival (EFS) and treatment effect (original hazard ratio [HR] 0.64, 95%-confidence interval [CI] 0.47-0.87, p=0.004; with synthetic control HR 0.66, 95%-CI 0.48-0.90, p=0.009). Our findings demonstrate that AI-generated synthetic patients can serve as statistically rigorous controls supporting novel trial designs.

Background: Changes in the extracellular matrix (ECM) are a recognised feature of aggressive prostate cancer, but they are not exploited in clinical decision-making. We aimed to develop automated quantitative ECM parameters to facilitate risk stratification for localised prostate cancer. Methods: 378 quantitative ECM parameters were derived from picrosirius red-stained diagnostic prostate biopsies in a cohort of 422 patients, matched 1:1 for recurrence, recruited to the CHHiP (Conventional or Hypofractionated High Dose Intensity Modulated Radiotherapy in Prostate Cancer) trial of radiotherapy fractionation for localised prostate cancer. These ECM parameters comprehensively described fibre architecture, gaps and ECM texture. Machine learning models at the level of both individual image tiles and patients defined how ECM parameters related to tumour versus normal prostate, Gleason grade group and recurrence. Shapley analysis was used to interpret ECM feature importance and develop signatures associated with recurrence. Results: Specific ECM patterns identified tumour versus normal prostate, Gleason pattern 4 versus 3 and recurrence. ECM patterns associated with recurrence were enriched in Gleason 4+3 patients, versus Gleason 3+4 patients. Shapley analysis revealed that biopsies from patients with recurrence had smaller more elongated gaps between fibres, with finer grained ECM texture and lower ECM homogeneity than less recurrent regions. Interpretation: Quantitative automated analysis of ECM architecture can inform probability of prostate cancer recurrence after radiotherapy; Features relating to ECM gap size and texture are of particular relevance.

Background: Glioblastoma remains resistant to most immune-based therapies. Surgery may create a perioperative window in which systemic immune activation and tumor antigen release intersect. We evaluated whether COVID-19 vaccination shortly before first glioblastoma surgery was associated with survival. Methods: We performed a retrospective single-center cohort study of adults with newly diagnosed glioblastoma undergoing initial biopsy or resection from 2021 to 2025. The primary exposure was documented COVID-19 vaccination within 100 days before first tumor surgery. Overall survival was analyzed from surgery using Kaplan-Meier and Cox models, with 1:1 propensity matching and sensitivity analyses addressing treatment completion, calendar time, surgical selection, steroid exposure, immune-cell variables, COVID severity, and negative-control vaccination. Results: The cohort included 187 patients: 64 perioperatively vaccinated and 123 non-perioperative comparators. Among vaccinated patients, 59/64 (92.2%) received mRNA vaccines; median vaccination-to-surgery interval was 81 days (IQR 71-90). Median overall survival was 743 days in vaccinated patients versus 318 days in comparators (unmatched HR 0.48, 95% CI 0.30-0.76; p=0.002). After 1:1 matching, median survival was 743 versus 349 days (HR 0.52, 95% CI 0.34-0.80). Sensitivity analyses accounting for adjuvant therapy, surgery year, extent of resection, steroid exposure, immune-cell measures, and COVID hospitalization were directionally consistent. Influenza vaccination was not associated with survival. Conclusions: COVID-19 vaccination within 100 days before first glioblastoma surgery was associated with longer overall survival. These findings identify perioperative vaccination timing as a potentially relevant and modifiable variable in glioblastoma outcomes.

Tumor-infiltrating lymphocytes (TILs) are widely used to assess antitumor immunity in breast cancer but may not reflect the functional competence of adaptive immune responses. We show that immune organization, reflected by tertiary lymphoid structures (TLS) and coordinated humoral and cellular immune programs, represents a distinct dimension of tumor immunity beyond lymphocyte abundance. By integrating histologic, transcriptomic, spatial, and immune receptor profiling analyses across multiple breast cancer cohorts, we show that immune organization is associated with greater immune repertoire diversity, evidence of therapy-induced clonal selection, and improved clinical outcomes, independent of immune infiltration. Transcriptomic measures of immune organization retained independent prognostic value across external cohorts, whereas measures of immune infiltration did not. Furthermore, treatment-induced increases in immune organization, but not immune infiltration, were associated with therapeutic response. These findings identify immune organization as a dynamic and clinically measurable state of adaptive antitumor immunity with implications for prognosis, treatment monitoring, and therapeutic development in breast cancer.

Background The faecal immunochemical test (FIT) is central to triaging symptomatic patients with suspected colorectal cancer (CRC) in UK primary care, yet only about one in eleven patients above the NICE 10 ug/g threshold have CRC. Existing prediction models attempting to improve on FIT have relied on conventional statistics and limited predictors. Methods GP-requested FITs with linked data (Jan 2017 - May 2025) were extracted from the Oxford University Hospitals (OUH) datawarehouse. Patients aged [&ge;]18 with core bloods and 180-day CRC follow-up were included. Machine learning (ML) models were trained on up to 1,025 predictors: FIT, age, sex, blood tests and their time series slopes, diagnoses/procedures/prescriptions, deprivation, BMI, and ethnicity. Models comprised penalised logistic regression, generalised additive models (EBM, NAM, SNAM, NODE-GAM), decision tree ensembles (random forests, XGBoost), and a multilayer perceptron. Referral reduction versus FIT [&ge;]10 ug/g was evaluated at model risk score thresholds capturing the same cancers (conservative) or same proportion of cancers (less conservative) as FIT. Potential to prioritise referred patients was assessed by examining whether positive predictive value (PPV) is very high (>30%) at any substantial sensitivity (>10%). Nested twice-repeated five-fold cross-validation provided unbiased estimates. An existing COLOFIT model was evaluated alongside. Findings 62,219 individuals (746 CRC) were analysed; 30,862 patients (315 CRC) with high/low risk symptoms and buffered FITs formed the primary subset. At [&ge;]10 ug/g, FIT had 91.4% sensitivity, 84.2% specificity, 5.6% PPV, and 99.9% NPV. No model reduced referrals when required to capture the same cancers as in the FIT [&ge;]10 ug/g cohort. Generalised additive models achieved up to 18.5% referral reduction when detecting the same proportion but some different cancers as FIT [&ge;]10 ug/g (EBM: 18.5%, NODE-GAM: 17.5%, SNAM: 17.4%, COLOFIT: 16.7%). At 30% sensitivity, EBM, NAM and NODE-GAM had average PPVs between 34.6%-35.0%, while FIT had a PPV of 14.6%. Interpretation Generalised additive models (GAMs) reduced referrals on average by 19% if a small proportion of the FIT-positive CRCs were substituted with originally FIT-negative CRCs by the models. No model, including COLOFIT, reduced referrals while capturing all FIT-positive cancers. Generalised additive models could detect about a third of CRCs faster, as one in three patients flagged by the models had CRC at 30% sensitivity. Funding EPSRC Centre for Doctoral Training in Health Data Science; National Institute for Health Research (NIHR) Oxford Biomedical Research Centre; Cancer Research UK. Keywords Colorectal cancer, faecal immunochemical test, machine learning, positive predictive value

We present FootNet, a 453-image multi-view smartphone foot dataset for binary foot segmentation, with expertannotated masks across six anatomical views (dorsal, medial, and plantar, both left and right). We benchmark four segmentation models under a controlled protocol: U-Net with a MobileNetV2 encoder achieves the best performance (IoU 0.9268, Dice 0.9608, 95 % CI [0.9209, 0.9320]); DeepLabV3 with MobileNetV3-Large scores IoU 0.8984 (Dice 0.9449); UNet++ with MobileNetV2 scores IoU 0.8913 (Dice 0.9391); and SAM ViT-B with oracle boundingbox prompt scores IoU 0.9219 on the matched 191-image subset. Bonferroni-corrected Wilcoxon signed-rank tests (k = 6 comparisons) show U-Net significantly outperforms DeepLab (p < 0.001, r = 0.638) and SAM ViT-B with oracle boundingbox (p = 0.005, r = 0.202); UNet++ does not significantly differ from DeepLab (p = 0.062). Connected-component postprocessing yields negligible benefit (mean {triangleup}IoU = +0.0003, 12 of 453 images improved). The extended dataset is available upon request

As machine learning and artificial intelligence systems are increasingly used in healthcare, rigorous evaluation of their classification performance has become critical. The F1 and F{beta} scores are widely adopted metrics for assessing performance in imbalanced biomedical data. Recently, we introduced psF1, a unified statistical framework for inference and study design for single and comparative F1 and F{beta} scores under the assumption of independent classifiers. In practice, however, benchmarking two classifiers on the same dataset creates a correlated paired setting. Ignoring this intrinsic dependency leads to overestimation of the standard error and a substantial loss of statistical power. To address this, we develop psF1pair, an advanced framework for statistical inference and power analysis that explicitly accounts for correlations between classifier pairs. Extensive simulation studies demonstrate the performance of psF1pair, and its utility is further illustrated through application to a real-world imaging classification system. As expected, higher correlation between classifiers yields narrower confidence intervals and enhanced statistical power. A freely available R package is provided to facilitate implementation, supporting accurate evaluation and study design for predictive and classification models in biomedical research.

Background. Patients on GLP-1 medications lose very different amounts of weight, and most published prediction models include only patients who complete six months. That design omits everyone who disengages earlier, which is the majority of the cohort. We built a tool that includes patients who disengage and delivers useful predictions at the week-8 visit, where the clinical decision is actually made. Methods. Beginning with 237,800 adults enrolled in a US telehealth GLP-1 program, we required a documented week-8 weight, a refill-confirmed dose, and reported ethnicity, yielding an analytic cohort of 22,538. We answered three questions: the patient's likely six-month weight loss and our confidence in it; the probability of dropout before six months; and when weight loss plateaus. For the first, we fit a cubic in week-8 percent loss plus 16 covariates, with quantile-regression bands at the 10th and 90th percentiles for the prediction interval, checking fractional-logit and isotonic recalibration as alternatives. For the second, we fit a logistic regression and compared it to gradient boosting. For the third, we fit a per-patient exponential trajectory among patients with at least four weight observations. We trained on enrollments before 2024-07-01 and tested on later ones, compared completer outcomes to published RCTs, and tested the week-8 anchor against measurements at weeks 2, 4, 6, 8, 10, 12, 16, and 20. Results. Mean six-month weight loss in completers was 11.7% on semaglutide and 14.1% on tirzepatide, in line with STEP-1 and SURMOUNT-1. Six-month disengagement was 66%. The prediction model reached test R2 = 0.65 with a mean absolute error of 2.76 percentage points. Calibration was strong: calibration-in-the-large was -0.52 pp and the calibration slope was 0.96. The 80% quantile-regression interval covered 76% of test patients; the 95% interval covered 93%. The disengagement model reached test AUC 0.79, against 0.74 for gradient boosting. Median plateau time among engaged patients was 387 days, longer in lower-BMI tertiles. The week-8 anchor gave R2 = 0.65, compared to 0.48 to 0.61 at earlier weeks and 0.67 to 0.91 at later weeks. We chose week 8 because 80% of slow responders reach their post-titration decision point at or before that visit. Two of twenty subgroup cells had reduced predictive accuracy; two more were too sparse to validate. Conclusions. Observed week-8 weight loss is the strongest predictor of six-month outcome. The model's accuracy (R2 = 0.65, MAE 2.76 pp) is appropriate for calibrating expectations and identifying patients for the post-titration decision, but not precise enough to drive that decision on its own. Disengagement is predictable at week 8 with AUC 0.79. Engaged patients plateau at a median of 387 days. Week 8 is the earliest visit at which titration is mostly complete, accuracy is in a useful range, and the post-titration decision remains actionable; later anchors predict better but inform a decision that has already been made for most patients. The model is temporally (internally) validated but not yet externally validated, and because it was developed on a single platform it should be regarded as a recalibration target rather than a drop-in deployment elsewhere. The tool is published as a public web calculator to support shared decision-making, though it is not precise enough on its own to drive an irreversible clinical decision. It is prognostic, not therapeutic; treatment-effect estimation is addressed in companion work.

The Cancer Genome Atlas (TCGA) holds clinical data for over 11,000 patients across 33 cancer types, but access is hard because of complex file structures, heterogeneous formats, and the need for programming. We present an agentic system for natural language querying and statistical analysis of TCGA clinical data. The system uses a large language model as an autonomous ReAct agent that selects from eight computational tools, including data extraction, descriptive statistics, Kaplan-Meier survival analysis with log-rank tests, hypothesis testing, and verification against the curated TCGA Pan-Cancer Clinical Data Resource (CDR). The agent reasons about intermediate results, adapts its approach, and returns clinically contextualized responses with source attribution and auditable traces. We introduce TCGA-Agent-Bench, 440 queries across five difficulty tiers with ground truth from the independently curated TCGA-CDR, evaluated with dual metrics of numerical accuracy and clinical completeness. The system achieves 93.4% overall accuracy (100% single-patient lookups, 99.1% cohort statistics, 92.8% comparative analyses), outperforming a fixed rule-based pipeline (87.1%), a single-pass LLM (81.8%), and retrieval-augmented generation (66.9% on a subset). Most of the benchmark is answerable from the CDR alone, so we locate the extraction layer's value in fields the CDR lacks (drug treatments, TNM components, biomarkers, biospecimen metadata): on 26 queries targeting these, the full system answers 100% versus 3.8% for CDR-only. Ablations show the reasoning loop is most impactful (+9.1% accuracy, +22.0 completeness points). A tool-based agentic architecture enables accurate, auditable analysis of clinical repositories, with value driven by tool design and recovered fields rather than model scale.

ABSTRACT Background: Despite its high prevalence and established impact on women's health, the long-term biological effects of Intimate Partner Violence (IPV) remain poorly understood. In particular, its potential role in increasing cancer risk has received limited attention. This review examines whether IPV may be associated with elevated cancer risk in women. Methods: We conducted a systematic review and meta-analysis in accordance with PRISMA and MOOSE guidelines to evaluate whether IPV may be associated with cancer risk. Eligible studies included adult women ([&ge;]18 years) with documented IPV exposure and cancer or precancerous outcomes. We searched PubMed, Web of Science, Scopus, and Google Scholar for articles published from 2000 to 2025. Study quality was assessed using the Newcastle-Ottawa Scale (NOS). A random-effects meta-analysis was performed on longitudinal studies reporting adjusted risk estimates. Results: Thirteen studies were included in the qualitative synthesis, but only two met criteria for meta-analysis, both reporting on cervical cancer. The pooled odds ratio was 3.00 (95% CI: 2.05 - 4.38; I2 = 0%). A separate pooled prevalence analysis of six retrospective studies showed that 32.2% of women with cancer reported a lifetime history of IPV. Study quality ranged from low to high. Conclusions: This review underscores the limited and heterogeneous nature of the existing evidence on IPV as a potential cancer risk factor. While preliminary findings suggest a possible association, particularly with cervical cancer, the scarcity of high-quality longitudinal studies and the methodological variability in the studies reviewed prevent definitive conclusions regarding causal linkage. Further research, particularly prospective and mechanistic studies, is needed to clarify the relationship between IPV and oncogenesis across different cancer types and to identify underlying biological pathways.

Objective To develop and validate a risk model for predicting postoperative bleeding in patients with thyroid cancer. Methods A total of 2800 consecutive patients diagnosed with thyroid cancer in the Department of Thyroid and Breast Surgery of the Affiliated Hospital of Xuzhou Medical University between January 2020 and December 2023 were retrospectively analyzed. Patients were categorized into two groups based on postoperative bleeding occurrence: bleeding and non-bleeding groups. Univariate and multivariate logistic regression analyses were utilized to screen independent risk factors. Meanwhile, risk prediction models were developed and nomogram . Subgroup analysis was performed to identify independent risk factors. The predictive effects of the models were assessed using the Hosmer-Lemeshow test and receiver operating characteristic (ROC) curves. Results Of the 2800 recruited patients, 50 had postoperative bleeding, with an incidence rate of 1.7%. Multivariate logistic regression analysis showed that age, hypertension, total thyroidectomy, tumor size [&ge;]4 cm, and operation time [&ge;]90 min were the risk factors for postoperative bleeding in thyroid cancer patients (P<0.05). A risk prediction model was established based on the above factors, and the area under the ROC curve was 0.881, with a sensitivity of 94.0%, a specificity of 67.3%, and an accuracy of 74.0%. Decision curve analysis revealed that the model had good predictive ability. Conclusions The constructed risk prediction model has good predictive power and can provide a reference for healthcare professionals to predict the risk of bleeding in patients after thyroid cancer surgery.

Aim: To assess temporal trends in incidence and mortality and project the future burden of five major gastrointestinal cancers in Jiangsu Province, China. Methods: Population-based cancer registry data from Jiangsu Province between 2010 and 2021 were used to analyze the burden of esophageal, gastric, colon, rectal, and liver cancers. Age-standardized incidence and mortality rates were calculated and compared by cancer type, sex, and urban-rural residence. Joinpoint regression was used to estimate annual percentage changes (APC) and average annual percentage changes (AAPC). The APC from the most recent Joinpoint segment was used to project incidence and mortality rates to 2030. Results: In 2021, gastric cancer had the highest age-standardized incidence and mortality among the five cancers. Incidence and mortality were consistently higher in males than in females and increased markedly after 50 years of age. From 2010 to 2021, age-standardized incidence and mortality declined for esophageal, gastric, and liver cancer, but increased for colon and rectal cancer. Colon cancer showed the steepest increase in both incidence and mortality. Rural areas experienced faster increases in colon and rectal cancer burden than urban areas. Projections to 2030 suggest continued declines in esophageal, gastric, and liver cancer, while colon cancer incidence and mortality are expected to rise further. Conclusion: Jiangsu Province is experiencing a transition in gastrointestinal cancer burden, with continued declines in esophageal, gastric, and liver cancers but an emerging and growing burden of colorectal cancer, especially colon cancer. Prevention strategies should focus on expanding colorectal cancer screening and early diagnosis, particularly in rural areas, while sustaining control of esophageal, gastric, and liver cancers.

This study presents ReCo (Research Cosmos), a self-configuring and self-extending agentic research framework for the biomedical domain. ReCo is orchestrated by a large language model that interacts with native computing tools, bundled Model Context Protocol (MCP) servers, structured skills, persistent project memory, and a desktop interface. Its bundled MCP servers provide biomedical analysis capabilities while serving as implementation paradigms for integrating new computational and AI frameworks. Structured skills encode procedures for environment configuration and framework ingestion, enabling ReCo to inspect repositories, manuscripts, or local codebases; identify dependencies and execution patterns; create isolated runtime environments; design and implement MCP interfaces. Self-extension was evaluated using five heterogeneous systems: the Merlin computed tomography foundation model, MAISI-v2 medical image synthesis framework, asari liquid chromatography-mass spectrometry workflow, DosimeTron agentic radiation-dosimetry platform, and Orthanc DICOM server. ReCo successfully operationalized all five systems and completed predefined functional evaluations. Re-hosted DosimeTron outputs demonstrated near-perfect agreement with the reference pipeline across 651 organ observations (Pearson correlation and Lin concordance correlation coefficient, 0.99999; mean absolute percentage difference, 0.37%). Notably, ReCo configured Orthanc as a PACS-like coordination layer, integrated it with DosimeTron, Merlin, and TotalSegmentator, and orchestrated data retrieval, analysis, and return of valid DICOM RTSTRUCT, RTDOSE, and Structured Report. ReCo provides a unified environment for configuring, documenting, and operationalizing heterogeneous biomedical frameworks, reducing technical barriers to the adoption and integration of emerging computational and AI methods. The official open-source ReCo GitHub repository is available at: https://github.com/eltzanis/ReCo

Background: Early detection of pancreatic ductal adenocarcinoma (PDAC) is critical. While faecal elastase-1 (FE-1) is a standard clinical marker for pancreatic function, its diagnostic accuracy for malignancy is limited. We sought to identify plasma metabolites that enhance FE-1 performance in symptomatic "at-risk" patients. Methods: Using the DEPEND cohort (CRUK C45617/A29908), plasma metabolomics was performed on patients with resectable PDAC (n=23) and healthy volunteers (n=24). Predictive modelling included feature selection and cross-validation, with further validation in an independent external cohort. Results: Citrulline was identified as significantly depleted in PDAC patients across discovery and validation cohorts. In isolation, Citrulline achieved an AUC of 0.86 (internal) and 0.88 (external validation). Standalone FE-1 demonstrated an AUC of 0.67. However, combining Citrulline and FE-1 significantly improved diagnostic performance, achieving a combined AUC of 0.96. Stratification revealed distinct metabolomic signatures associated with poorly differentiated tumours, suggesting a link to histological grade. Conclusions: Integrating Citrulline with FE-1 testing substantially improves PDAC detection in symptomatic patients. This non-invasive panel offers high diagnostic potential, though prospective validation is required to establish clinical cut-offs for routine practice.