Prognostic stratification of Hodgkin lymphoma (HL) patients in ICU remains challenging, with conventional scoring systems often overlooking pathophysiological biomarkers. This retrospective cohort study analyzed 1,908 HL patients from the MIMIC-IV database. Multivariate logistic regression and machine learning (ML, gradient-boosting (GBM) was optimized with LASSO regularization) were employed to identify 30-day mortality predictors, validated through SHAP interpretability, calibration curves, and decision curve analysis. Multi-organ dysfunction (AST, BUN, T-Bil), systemic inflammation (NLR, WBC) and APTT emerged as critical mortality determinants, and selected for model construction. GBM achieved superior discrimination (training AUC = 0.89; test AUC = 0.75), SHAP analysis, calibration curve and decision curve analysis (DCA) confirmed clinical utility, outperforming empirical intervention strategies. This study establishes a biomarker-driven ML framework for HL prognosis, integrating renal, hepatic, and inflammatory markers into actionable risk stratification. thereby providing a scientific basis for comprehensive HL management.

Brain invasion is an independent diagnostic criterion for WHO grade 2 meningiomas, and preoperative prediction of brain invasion in meningiomas is crucial for guiding treatment decisions. Therefore, we constructed a radiomics model that integrated structural and diffusion-weighted images to predict brain invasion of meningiomas. Seven hundred and twenty-three consecutive patients with pathologically confirmed meningiomas between 2013 and 2022 were retrospectively studied. Radiomics features of the brain-to-tumor interface region were extracted from structural MRI and DWI-derived apparent diffusion coefficient (ADC) maps. The least absolute shrinkage and selection operator (LASSO) method was utilized to select radiomics features. A linear predictor of brain invasion was constructed using a logistic regression classifier. The model's performance was evaluated using receiver operating characteristic (ROC) curve analysis. Additionally, decision curve analysis (DCA) was performed to evaluate the clinical utility of the established models. A nomogram was developed for a combined model that incorporates clinical features, along with radiomics scores derived from structural images and ADC maps. DeLong test and integrated discrimination improvement (IDI) were used to compare the diagnostic efficiency of different models. Six radiomics features from structural MRI, six radiomics features from ADC, the volume of peritumoral edema, and gender were selected to construct the combined model. This model achieved the highest AUC and sensitivity for predicting brain invasion in both the training (AUC = 0.897, 95%CI: 0.857 to 0.936, sensitivity = 0.911) and test sets (AUC = 0.871, 95%CI: 0.806 to 0.936, sensitivity = 0.895). It outperformed the structural model (AUC = 0.691) and the structural and clinical model (AUC = 0.812). The IDI demonstrated a significant improvement in predictive value when ADC radiomic features were added to the combined model. The incorporation of ADC radiomics into the MRI radiomic model improved the diagnostic performance for identifying brain invasion in meningiomas.

Background Granulomatous lobular mastitis (GLM) frequently mimics ductal carcinoma in situ (DCIS) in clinical presentation and imaging characteristics, leading to misdiagnosis and unnecessary aggressive interventions. This study aimed to develop and validate a practical nomogram for differentiating GLM from DCIS. Methods We conducted a retrospective study at Quanzhou First Hospital from January 2020 to April 2025, including 290 patients with histopathologically confirmed GLM (n=128) or DCIS (n=162). Patients were randomly divided into training (n=203) and validation (n=87) sets. Clinical, laboratory, and ultrasound features were analyzed using univariate and multivariate logistic regression to identify independent predictors. A nomogram was constructed and evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis. Results Six independent predictors were incorporated into the final nomogram: age, lesion size, margin characteristics, microcalcifications, posterior acoustic enhancement, and peri-lesional flow. The nomogram demonstrated excellent discriminative performance with areas under the ROC curve of 0.95 (95% CI: 0.92-0.98) in the training set and 0.93 (95% CI: 0.88-0.98) in the validation set. At optimal thresholds, the model achieved sensitivity of 92% and specificity of 89% in the training set, and 89% and 79% respectively in the validation set. Calibration plots confirmed high predictive accuracy, and decision curve analysis demonstrated substantial clinical benefit across clinically relevant threshold probabilities. Conclusions This novel nomogram represents a diagnostic tool specifically designed for GLM versus DCIS differentiation. Its reliance on widely available clinical and ultrasound parameters makes it particularly valuable for resource-limited settings, potentially reducing unnecessary biopsies and associated patient morbidity.

Introduction Parkinson’s disease (PD) is the second most common neurodegenerative disorder. The risk of frailty is significantly higher in patients with PD than in age-matched individuals without PD. This study aimed to develop a machine learning–based predictive model for frailty in PD. Methods We conducted a cross-sectional study of early- and middle-stage PD patients recruited from June 2024 to June 2025 at Shenzhen People’s Hospital. Frailty was assessed using the Fried criteria (five components: gait speed, grip strength, physical activity, fatigue, and weight loss). A total of 42 demographic and clinical variables, including disease history, Montreal cognitive assessment (MoCA), and unified Parkinson’s disease rating scale (MDS-UPDRS) scores, were collected and compared between PD patients with and without frailty. Spearman correlation and LASSO regression were used to identify independent risk factors. Multiple machine learning algorithms were applied to construct predictive models. Model performance was evaluated using receiver operating characteristic (ROC) curves, area under the ROC curve (AUC), decision curve analysis (DCA), calibration plots, and forest plots. Results A total of 205 PD patients were enrolled (133 non-frail, 72 frail; mean age non-frail 62.92 ± 9.69 years, frail 68.13 ± 8.44 years). Significant group differences were found in sex ( p = 0.013), age ( p < 0.001), disease severity (MDS-UPDRS, p < 0.001; modified Hoehn-Yahr stage (H&Y stage), p < 0.001), alcohol consumption ( p = 0.010), MoCA ( p < 0.001), HAMD ( p = 0.001), and Hamilton anxiety rating scale (HAMA) ( p < 0.001). Eight features were identified as independent predictors of frailty: sex, age, alcohol use, Modified H&Y stage, UPDRS-IV score, HAMA score, executive function, and naming. Among all tested algorithms, logistic regression achieved the best predictive performance (AUC = 0.83 in the test set), outperforming other machine learning models. Conclusion Frailty in PD was associated with female sex, older age, alcohol use, and more advanced disease severity. Patients with PD and frailty exhibited higher MDS-UPDRS scores, more severe cognitive impairment, and greater levels of depression and anxiety. Integrating clinical data with machine learning, especially logistic regression, provides a reliable and scalable tool for early identification and risk stratification of frailty in PD.

Stage B heart failure (HF) requires precise risk stratification. Current guideline-recommended approaches focus on cardiac indicators, neglecting functional status. The Short Physical Performance Battery (SPPB) predicts outcomes in symptomatic HF, but its prognostic value for mortality in older adults with Stage B HF remains unexplored. This study aimed to evaluate this association, establish an optimal SPPB threshold, and determine the incremental value of SPPB and its subtests over conventional metrics. This prospective cohort enrolled patients with Stage B HF aged ≥ 65years and followed them for 5years to evaluate all-cause mortality. Restricted cubic splines were applied to explore potential non-linear relationships between SPPB and mortality. Prognostic thresholds were identified using segmented regression. Associations with mortality were examined using multivariable Cox regression. Incremental predictive value of SPPB and its subtests over the base model was evaluated using area under the curve (AUC), integrated discrimination improvement (IDI), continuous net reclassification improvement (cNRI), and decision curve analysis (DCA). A total of 527 patients were included (mean age 75.3 ± 6.4years; 53.9% female). SPPB showed an L-shaped association with all-cause mortality, with a threshold of 6. After adjustment for age, sex, log-transformed NT-proBNP, and left atrial anteroposterior diameter, SPPB ≤ 6 was associated with a threefold higher risk of all-cause mortality (HR = 3.29, 95% CI: 1.90-5.70) compared with SPPB > 6. Using the 6-point threshold of the SPPB, rather than the conventional 9, more effectively improved the predictive performance (AUC = 0.840; IDI = 0.061; cNRI = 0.433). DCA further showed higher net benefit for the 6-point model across threshold probabilities of 0.05-0.50. Among the subtests, gait speed provided the greatest incremental value (AUC = 0.833; IDI = 0.053; cNRI = 0.258). SPPB was independently associated with mortality in older patients with Stage B HF. In this dataset, a 6-point threshold showed improved discrimination for identifying higher-risk individuals compared with the conventional 9-point threshold. Gait speed showed the strongest incremental prognostic value among the subtests. Incorporating SPPB into prognostic assessment may improve risk stratification in this population. ChiCTR1800017204; 07/18/2018.

Latent tuberculosis infection (LTBI) is a significant reservoir for active tuberculosis development. Identifying key risk factors is crucial for prevention strategies. Machine learning techniques can uncover complex relationships between risk factors and disease outcomes. Data were collected from China's Tuberculosis Management Information System. LTBI was defined by positive tuberculin skin tests. A case-control design comparing LTBI (n = 669) with active tuberculosis (ATB, n = 669) patients was employed. Propensity score matching (1:1) was performed using age, gender, and education level. Four machine learning models (random forest, XGBoost, support vector machine, and neural network) were developed for feature importance analysis. Least Absolute Shrinkage and Selection Operator (LASSO) regression and logistic regression identified key risk factors. Bootstrap resampling (n = 1,000 iterations) assessed model stability with 95% confidence intervals. Shapley Additive Explanations (SHAP) analysis provided feature importance interpretation. A risk nomogram was constructed and evaluated using receiver operating characteristic curves, calibration plots, and decision curve analysis. Among 1,338 matched participants, XGBoost demonstrated superior performance (AUC = 0.898, accuracy = 85.7%, sensitivity = 84.2%, specificity = 86.9%). SHAP analysis revealed age group (mean |SHAP value|=0.818) as the most influential predictor, followed by medical insurance type (0.599), income group (0.523), and education level (0.439). Logistic regression identified 11 significant risk factors: age (OR = 2.35, 95%CI: 1.86-2.96), BMI (OR = 0.81, 95%CI: 0.71-0.93), smoking status, occupational dust exposure, diabetes, medical insurance type, immunosuppressant use, education level, silicosis, anemia, and TB contact history. The nomogram showed good discrimination (AUC = 0.839) and clinical utility, identifying 64.44% of subjects as high-risk with 53.62% confirmed as true positives at 20% risk threshold. This study successfully identified key LTBI risk factors using machine learning approaches. The developed nomogram provides a practical tool for targeted screening in resource-limited settings. Interventions targeting modifiable factors such as smoking cessation and occupational dust control may reduce LTBI and active TB burden.

Hypoxemia is a common complication of bronchoscopy performed under deep sedation in pediatric patients, seriously compromising the safety of surgery and the prognosis of children. Therefore, this study explored the risk factors and established a predictive model for hypoxia during bronchoscopy in pediatric patients under deep sedation. 365 pediatric patients who underwent bronchoscopy under deep sedation in our hospital from January to December 2024 were retrospectively selected with a random number table. After screening, 346 pediatric patients were finally included, and they were divided into a modeling group (n = 243) and a validation group (n = 103) in a ratio of 7:3. Data were analyzed. The results of binary logistic regression analysis showed that age (6.39 ± 2.80) and examination duration were factors influencing hypoxemia during bronchoscopy under deep sedation (P < 0.05). A predictive model was developed. The calibration curves in both the modeling group and validation group showed lines close to a slope of 1, indicating good consistency between the predicted risks and the actual risks. The ROC (receiver operating characteristic) analysis results showed that the area under the curve in the modeling group was 0.96. In the validation group, the area under the curve was 0.89. The DCA (decision curve analysis) curve demonstrated a clear net benefit of the model. Given that young age (6.39 ± 2.80) and long examination duration are important risk factors for hypoxia during bronchoscopy under deep sedation in pediatric patients, preoperative assessment of age and optimization of the procedure to reduce its duration are recommended. At the same time, based on the verified prediction model, high-risk children should take measures to prevent hypoxia in advance.

Development of a machine learning-based prediction model for postoperative delirium in frail elderly patients undergoing noncardiac surgery under general anesthesia.

To develop and internally validate a nomogram for predicting the likelihood of depression among adult women with urinary incontinence (UI) using data from a nationally representative survey. This study included 6308 women with UI aged 20 years or older from the National Health and Nutrition Examination Survey (NHANES) 2007-2018. The women were selected at random: 75% were the training set and the remaining 25% comprised the testing set. Least absolute shrinkage and selection operator (LASSO) binomial and logistic regression models were used to select the optimal predictive variables. The depression probability was calculated using a predictor-based nomogram. Receiver operating characteristics area under the curve (ROC-AUC), calibration curve, and decision curve analysis (DCA) were used to evaluate the nomogram's performance. The nomogram included 11 predictors: age, education, ratio of family income to poverty, smoking, stroke, sleep time, trouble sleeping, leakage bother level, daily activities affected, number of nighttime urinations, and moderate-vigorous recreational activity. A nomogram model for depression risk was established based on these predictors. The AUC of the nomogram was 0.811 (95% confidence interval [CI] 0.793-0.829) in the training set and 0.810 (95% CI 0.780-0.839) in the testing set. The nomogram was well calibrated according to the calibration curve. The DCA demonstrated that the nomogram was clinically useful. This study established a nomogram that is helpful for screening indivudals with UI at high risk of depression and assisting gynecologists in identifying signs of depression in patients and providing treatment.

Acute kidney injury (AKI) is a prevalent and severe complication following non-cardiac surgery, often leading to poor outcomes. Despite the critical role of inflammation in AKI pathogenesis, reliable preoperative predictive models remain elusive. The pan-immune inflammation value (PIV), a novel index that integrates counts of neutrophils, platelets, lymphocytes, and monocytes, provides a comprehensive reflection of systemic inflammation. This study aimed to develop and validate a clinical prediction model for postoperative AKI (PO-AKI) in non-cardiac surgical patients. This retrospective study included adult patients who underwent non-cardiac surgery under general anaesthesia. The objective was to construct a model to predict PO-AKI. The statistical analysis focused on model construction and validation. LASSO regression was employed for variable selection to identify the most parsimonious set of predictors. The model's performance was evaluated based on its discriminative ability (AUC), with calibration and decision curve analysis used to assess its clinical utility. The cohort consisted of 1,164 adult patients. AKI was diagnosed in 8.4% of patients. The primary outcome, the performance of the prediction model, showed an AUC of 0.70. The model incorporated PIV and emergency surgery. The secondary outcome, the discriminative ability of PIV alone, yielded an AUC of 0.691. The model demonstrated good calibration and provided a clinical net benefit across a wide range of threshold probabilities. We developed and validated a prediction model for PO-AKI. This model, which integrates PIV and emergency surgery, serves as an effective tool for preoperative risk stratification, facilitating the identification of high-risk patients and optimizing perioperative management.

This study aimed to construct and validate a predictive model for fear of disease progression in patients after percutaneous coronary intervention(PCI). From March to October 2024, 455 post-PCI patients in the Department of Cardiovascular Medicine of a tertiary general hospital in Sichuan Province, China, were randomly divided into a training set and a validation set in a ratio of 7:3 as study subjects. LASSO regression and multifactorial logistic regression were used to analyze the factors influencing fear of disease progression in post-PCI patients, and a column chart was constructed. The predictive performance of the model was evaluated using the area under the ROC curve, Hosmer-Lemeshow test, and calibration curve. Clinical effectiveness was evaluated using clinical decision curve analysis. Among 455 post-PCI patients, 295 had a fear of disease progression, with an incidence of 64.8%. Seven influencing factors, including average monthly family income, number of chronic diseases, disease duration, number of interventional treatments, number of stent implants, psychological resilience, and perceived social support, were screened to construct the prediction model. The area under the ROC curve of the prediction model in the training set and the validation set were 0.941 (95% CI: 0.915-0.967) and 0.947 (95% CI: 0.911-0.984), respectively; the results of the Hosmer-Lemeshow goodness-of-fit test were χ2 = 12.564 (P = 0.128) and χ2 = 3.758 (P = 0.878); calibration curves showed significant agreement between predicted and actual values. The clinical decision curve analysis demonstrates that this model exhibits favorable net benefit and clinical effectiveness.The fear of disease progression prediction model constructed in this study has good predictive ability, which can provide a reference basis for effectively identifying high-risk groups and formulating targeted interventions to reduce the fear of disease progression in post-PCI patients.

This research seeks to formulate and confirm a deep learning radiomics nomogram (DLRN) based on nonenhanced CT (NECT) to differentiate intraparenchymal hematomas associated with Cerebral Venous Thrombosis (CVT) from those caused by other etiologies. 275 patients with intraparenchymal hematomas who underwent NECT were included in this work. Participants from two medical centers were assigned to distinct cohorts: a training set from Center 1 consisting of 192 patients (46 with confirmed CVT and 146 with other etiologies) and an external test set from Center 2 comprising 83 patients (24 with confirmed CVT and 59 with other etiologies). Conventional radiomics (Rad) features and deep learning (DL) features were derived from NECT images and integrated to form deep learning radiomics (DLR) features. Separate predictive models were constructed using Rad, DL, and DLR features. A DLR signature was obtained and integrated with medical characteristic variables to develop the DLRN model via multivariate logistic regression. The model's predictive performance was evaluated using ROC curves and decision curve analysis (DCA). Sixteen Rad features and three DL features were selected to construct the fused DLR features. The DLR model exhibited superior discriminative performance in identifying secondary intraparenchymal hemorrhage due to CVT compared to the individual Rad and DL models, achieving an AUC of 0.904(95% CI: 0.8207-0.9879) in the external test cohort. By integrating the DLR signature with epilepsy, the DLRN model was developed, demonstrating the highest predictive accuracy among all radiomics models, with an AUC of 0.911(95% CI: 0.8265-0.9962) in the external test cohort. Decision curve analysis (DCA) revealed that the DLRN model offered enhanced practical applicability relative to other radiomics-based models. A CT-based DLRN model was developed to distinguish intraparenchymal hematomas associated with CVT and those caused by other factors. The model offers a rapid and non-invasive diagnostic approach without the need for contrast enhancement, potentially improving early diagnosis and clinical decision-making.

Robot-assisted partial nephrectomy (RAPN) is an established, minimally invasive technique to treat patients with renal masses. The incidence of acute kidney injury (AKI) after RAPN is high and is associated with poor prognosis. This study aims to develop and validate an interpretable machine-learning model based on clinical features for individualized risk assessment of RAPN-AKI. We retrospectively reviewed 325 patients undergoing RAPN at the Third Medical Center of PLA General Hospital (May 2022-Oct 2023) as the training dataset, and 146 from the Fifth Medical Center of PLA General Hospital (Nov 2023-Dec 2024) for external validation. Models were constructed using Boruta-selected features and eight machine learning algorithms. Performance was assessed by the area under the receiver operating characteristic curve (AUC), F1-score, accuracy, precision, calibration, and decision curve analysis (DCA). Shapley additive explanations (SHAP) interpreted feature contributions. The incidence of AKI in internal training and external validation datasets was 24.6% and 26%, respectively. The Boruta algorithm identified duration of renal artery blockade, preoperative serum creatinine (Scr), gender, body mass index (BMI), and age as important features. Among the eight machine learning models, the Gradient Boosting Machine (GBM) model demonstrated the best and most stable predictive outcomes in the internal training dataset (AUC = 0.889) and external validation dataset (AUC = 0.779). Both the calibration curve and DCA indicated better calibration and greater net benefit. SHAP analysis revealed the contribution of important features in the following order: duration of renal artery blockade, Scr, BMI, age, and gender. Dependency plots showed that duration of renal artery blockade > 22min, Scr > 80 µmol/L, BMI > 25kg/m², age > 60 years, and male were significantly associated with an increased risk of AKI. The GBM model exhibited strong predictive performance in both internal training dataset and external validation dataset and has the potential to assist clinicians to identify the high-risk patients early, enabling timely interventions that may reduce the incidence of RAPN-AKI and improving clinical outcomes. While, the interpretable machine learning model is currently applicable only to patients with low-risk or normal preoperative renal function.

BackgroundPreoperative diagnosis of microvascular invasion (MVI) is difficult for patients with hepatocellular carcinoma (HCC). The aim of this study was to develop and validate a nomogram to predict the risk of MVI before surgery.MethodsA total of 661 HCC patients who underwent curative resection were included in the study. Independent risk factors were identified by univariate/multivariate analyses and were built into a nomogram to estimate the risk of MVI. The receiver operating characteristic (ROC) curve, concordance index (c-index), calibration curve and decision curve analysis were used to evaluate the predictive performance of the models.ResultsPrealbumin, gamma-glutamyl transpeptidase, alpha-fetoprotein level, and tumor size were found to be independent risk factors for MVI and formed the basis of the nomogram. The area under the ROC curve (AUC) of the nomogram for predicting MVI was 0.775 (C-index of 0.781) in the training cohort, 0.787 (C-index of 0.785) in the validation cohort, and 0.789 (C-index of 0.790) in the external validation cohort. The nomogram exhibited favorable calibration performance, and decision curve analysis demonstrated that the nomogram has clinical value.ConclusionsWe developed a new nomogram that used basic clinical and laboratory variables to predict the probability of MVI before surgery for HCC patients. This nomogram can help clinicians choose appropriate treatment procedures.

Prognostic assessment of sepsis-induced acute respiratory distress syndrome in older patients using clinical and CT-based radiomic features.

Postoperative meningitis (PM) is a severe complication following skull base tumor surgery, often resulting in prolonged hospitalization and increased morbidity. However, predicting the duration of PM remains challenging. This study aimed to identify prognostic factors influencing the duration of PM and to develop a predictive nomogram to guide individualized management and antibiotic therapy. The authors conducted a retrospective cohort study of patients diagnosed with PM after skull base tumor surgery at a high-volume neurosurgical center between December 2018 and August 2024. Patients who received antibiotic treatment were included. Logistic regression analysis was performed to identify independent predictors of prolonged meningitis duration (> 7 and > 14 days). Based on these factors, predictive nomograms were developed and validated to estimate the probability of extended recovery times. Among 629 patients with PM, 240 (38%) experienced meningitis > 7 days and 69 (11%) had durations > 14 days. Multivariate analysis identified fever duration (> 38.5°C), highest CSF white blood cell (WBC) count, lowest CSF glucose level, highest blood WBC count, highest blood neutrophilic granulocyte proportion, repeat operation, and surgical approach as independent predictors of prolonged meningitis. The nomogram demonstrated good predictive performance, with concordance indices of 0.80 (95% CI 0.74-0.83) for the 7-day model and 0.79 (95% CI 0.70-0.85) for the 14-day model in the training cohort. Calibration curves and decision curve analyses further confirmed the accuracy and clinical utility of the models. The authors successfully developed and validated a prognostic nomogram to predict the duration of PM following skull base tumor surgery. This tool enables individualized risk stratification, informs the optimal duration of antibiotic therapy, and supports improved postoperative management. Prospective studies are warranted to further validate these findings across broader clinical settings.

BackgroundThe Kidney Failure Risk Equation (KFRE) is widely used for predicting kidney failure, but its external validity in Latin America is limited. A previous study in Peru found that KFRE was miscalibrated but did not evaluate its recalibration or clinical utility.MethodsWe conducted a retrospective cohort study using data from EsSalud’s Renal Health Surveillance Program (2013–2022), including 30,031 patients with chronic kidney disease (CKD) stages G3-4. Kidney failure was defined by dialysis initiation or nephrologist-confirmed end-stage renal disease. Calibration was assessed using observed-to-expected (O/E) ratios and differences, calibration slope, and intercept, while discrimination was evaluated using the concordance index (C-index). Recalibrated models were developed, and decision curve analysis (DCA) was performed to evaluate clinical utility.ResultsThe original KFRE demonstrated good discrimination (C-index: 0.88 at 2 years, 0.85 at 5 years) but poor calibration in-the-large: O/E ratios indicated mean underestimation of risk at 2 years (O/E ratio: 1.84) and a slight mean overestimation at 5 years (O/E ratio: 1.06). Original KFRE also had poor weak (slope: 0.58) and poor moderate calibration. Recalibrated models improved calibration in-the-large, but none achieved good weak (all slope < 1) and moderate calibration. However, DCA showed a higher net benefit for KFRE-based nephrology referrals (in original and recalibrated by method D) compared to Peruvian and international guidelines, especially over a 5-year horizon.ConclusionsDespite miscalibration, KFRE remains valuable for guiding nephrology referrals in Peru, with recalibrated models offering potential improvements. This is the first study in Latin America to rigorously assess the clinical utility of KFRE.Clinical trial numberNot applicable. This study is not a clinical trial.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12882-025-04357-z.

Amniotic fluid embolism (AFE) represents an uncommon yet life-threatening obstetric emergency characterized by its sudden onset and significant contribution to maternal death. Immune dysregulation has been implicated as a key pathological mechanism in AFE progression. This investigation employed through comprehensive bioinformatics approaches transcriptomic profiling of clinical specimens to systematically identify immune-associated genetic alterations that can distinguish AFE - induced pulmonary embolism from normal controls. We conducted RNA sequencing on clinical specimens to delineate differentially expressed genes associated with AFE pathogenesis and immune responses. Functional annotation and protein-protein interaction (PPI) networks were generated using specialized R packages. Machine learning algorithms facilitated the selection of candidate biomarkers, whose expression patterns were quantitatively assessed. Diagnostic performance was evaluated through nomogram construction, while immune cell infiltration patterns were characterized using computational deconvolution methods. Potential N6-methyladenosine (m6A) sites were predicted via established databases, and regulatory networks were reconstructed. Future validation will include quantitative Polymerase Chain Reaction (PCR) verification of critical gene expression. Our analysis identified two upregulated biomarkers (MMP9 and PPBP) in AFE samples. The constructed nomogram demonstrated that elevated biomarker scores correlated with increased AFE probability. Validation through calibration curves, decision curve analysis, and receiver operating characteristic curves confirmed robust predictive accuracy and clinical applicability. Immunological assessment revealed significant negative correlations between biomarker expression and central memory CD4 + T cells, activated CD8 + T cells, and activated dendritic cells. Both biomarkers exhibited moderate to high-confidence RNA methylation sites. We have provided valuable peripheral blood sequencing data from AFE patients and healthy controls. Through bioinformatics analysis, we identified two immune-related biomarkers: MMP9 and PPBP and systematically examines their biological significance through immune infiltration profiling, gene set enrichment, regulatory network construction, and pharmacological association studies. This study provides preliminary evidence for the diagnostic potential of these two markers for AFE and offers new insights for research into the immune response associated with AFE.

Purpose We develop an uncertainty-aware, Explainable Artificial Intelligence (XAI)-enabled probabilistic framework to predict and explain delay-driven cost risk in construction, accounting for time-varying exposures driven by supply reliability, regulatory cadence and labor stability. Design/methodology/approach Using data from 46 US high-rise projects, we estimate a hierarchical competing-risks Weibull survival model with time-varying covariates and project-level random effects via HMC/NUTS. We link posterior predictive delay exceedance to a stochastic cost overrun layer and integrate XAI through posterior-aware SHAP (global and local importances with 95% credible bands), interaction effects, and counterfactual recourse. Decision-curve analysis quantifies net benefit across operational trigger thresholds. Findings Relative to non-XAI baselines, the approach improves time-to-event discrimination and calibration (e.g. median C-index 0.81 vs 0.73; IBS reduction −0.027; both with 95% credible intervals). Global explanations identify supply reliability as the dominant driver, with a positive supply-regulation interaction. Scenario analyses show median reductions of 11–20% in cost overruns under feasible interventions (e.g. reliability uplift and buffer policies), with uncertainty reported. Practical implications The suggested framework provides a clear and interpretable information to the project managers both locally and globally. It recognizes the actual counterfactual activities that are associated with procurement scheduling and vertical logistics, it determines the decision-thresholds whose anticipated benefits are clearly outlined. All these characteristics allow more active, transparent and evidence-based management of complex project risks. Originality/value As far as we can determine, this paper is the first attempt to integrate hierarchical competing-risks Weibull modeling with uncertainty-aware explainable AI and a structural interdependence between schedule delay and cost escalation in a high-rise construction. The result is an interpretable, more practically oriented decision-support system that converts the findings of the analysis into practical directions to managers.

Objective This study aimed to investigate independent risk factors for clinically significant prostate cancer (csPCa) using serologic indices, multiparametric magnetic resonance imaging (mpMRI), and sound touch elastography (STE), and to develop and validate a nomogram-based prediction model using the optimal model derived from these factors. Methods A total of 240 patients who underwent ultrasound-guided transperineal prostate biopsy at Anqing Municipal Hospital between January 2024 and December 2024 were retrospectively enrolled. After applying exclusion criteria, 160 patients were included in the modeling cohort, which was divided into clinically significant prostate cancer (csPCa) and non-clinically significant prostate cancer (non-csPCa) groups based on pathological results. Additionally, 40 eligible patients from December 2024 to February 2025 were selected as the external validation cohort. Baseline data of the modeling cohort were collected, and independent risk factors for csPCa were identified using univariate and multivariate logistic regression analyses. The optimal model was selected by comparing with single-modal models, followed by developing a Nomogram prediction model. R language was used to plot decision curve analysis (DCA) for clinical utility evaluation, while receiver operating characteristic (ROC) curve and calibration curve were employed to assess predictive performance. Results Multivariate logistic regression analysis identified The Prostate Imaging Reporting and Data System score, age, free-to-total (f/t) prostate-specific antigen (PSA), Emax, TZ-ratio (transition zone ratio), and lesion density as independent risk factors for csPCa (all P &amp;lt; 0.05). The combined independent risk factor model demonstrated superior predictive performance compared to single-modal models, with an area under the receiver operating characteristic curve (AUC) of 0.926, sensitivity of 88.0%, and specificity of 83.1%. A nomogram model was developed based on this optimal model. Decision curve analysis (DCA) revealed substantial clinical benefit and high usability across a wide range of threshold probabilities. Calibration curve validation showed excellent predictive accuracy, with close agreement between predicted and observed probabilities. Both internal and external validation cohorts confirmed consistent predictive performance of the model. Conclusion The nomogram model integrating serologic indices, multiparametric mpMRI, and STE provides a more accurate and reliable tool for diagnosing csPCa, demonstrating substantial potential for clinical translation.