Search Results (521)

Background: Hip fractures are common in the elderly and often require ICU admission post-surgery due to high ASA scores and comorbidities. Length of hospital stay after ICU is a crucial indicator affecting patient recovery, complication rates, and healthcare costs. This study aimed to develop and validate a machine learning-based model to predict the factors associated with extended hospital stay (>7 days from surgery to discharge) in hip fracture patients requiring postoperative ICU care. The findings could help clinicians optimize ICU bed utilization and improve patient management strategies. Methods: In this retrospective single-centre cohort study conducted in a tertiary ICU in Turkey (2017–2024), 366 ICU-admitted hip fracture patients were analysed. Conventional statistical analyses were performed using SPSS 29, including Mann–Whitney U and chi-squared tests. To identify independent predictors associated with extended hospital stay, Least Absolute Shrinkage and Selection Operator (LASSO) regression was applied for variable selection, followed by multivariate binary logistic regression analysis. In addition, machine learning models (binary logistic regression, random forest (RF), extreme gradient boosting (XGBoost) and decision tree (DT)) were trained to predict the likelihood of extended hospital stay, defined as the total number of days from the date of surgery until hospital discharge, including both ICU and subsequent ward stay. Model performance was evaluated using AUROC, F1 score, accuracy, precision, recall, and Brier score. SHAP (SHapley Additive exPlanations) values were used to interpret feature contributions in the XGBoost model. Results: The XGBoost model showed the best performance, except for precision. The XGBoost model gave an AUROC of 0.80, precision of 0.67, recall of 0.92, F1 score of 0.78, accuracy of 0.71 and Brier score of 0.18. According to SHAP analysis, time from fracture to surgery, hypoalbuminaemia and ASA score were the variables that most affected the length of stay of hospitalisation. Conclusions: The developed machine learning model successfully classified hip fracture patients into short and extended hospital stay groups following postoperative intensive care. This classification model has the potential to aid in patient flow management, resource allocation, and clinical decision support. External validation will further strengthen its applicability across different settings. Full article

Imaging in controlled agriculture helps maximize plant growth by saving labor and optimizing resources. By monitoring specific plant traits, growers can prevent crop losses by correcting environmental conditions that lead to physiological disorders like leaf tipburn. This study aimed to identify morphometric and spectral markers for the early detection of tipburn in two Romaine lettuce (Lactuca sativa) cultivars (‘Chicarita’ and ‘Dragoon’) using an image-based system with color and multispectral cameras. By monitoring tipburn in treatments using melatonin, lettuce cultivars, and with and without supplemental lighting, we enhanced our system’s accuracy for high-resolution tipburn symptom identification. Canopy geometrical features varied between cultivars, with the more susceptible cultivar exhibiting higher compactness and extent values across time, regardless of lighting conditions. These traits were further used to compare simple linear, logistic, least absolute shrinkage and selection operator (LASSO) regression, and random forest models for predicting leaf fresh and dry weight. Random forest regression outperformed simpler models, reducing the percentage error for leaf fresh weight from ~34% (LASSO) to ~13% (RMSE: 34.14 g to 17.32 g). For leaf dry weight, the percentage error decreased from ~20% to ~12%, with an explained variance increase to 94%. Vegetation indices exhibited cultivar-specific responses to supplemental lighting. ‘Dragoon’ consistently had higher red-edge chlorophyll index (CIrededge), enhanced vegetation index, and normalized difference vegetation index values than ‘Chicarita’. Additionally, ‘Dragoon’ showed a distinct temporal trend in the photochemical reflectance index, which increased under supplemental lighting. This study highlights the potential of morphometric and spectral traits for early detection of tipburn susceptibility, optimizing cultivar-specific environmental management, and improving the accuracy of predictive modeling strategies. Full article

To address low training efficiency and poor reconstruction in traditional K Singular Value Decomposition (KSVD) for compressive sensing of shock wave signals, this study proposes an improved algorithm, DH-KSVD, integrating dynamic pruning and hybrid coding. The dynamic pruning mechanism eliminates redundant atoms according to their contributions and adaptive thresholds, while incorporating residual features to enhance dictionary compactness and training efficiency. The hybrid sparse constraint integrates the sparsity of ${\ell }_0$-Orthogonal Matching Pursuit (OMP) with the noise robustness of ${\ell }_1$-Least Absolute Shrinkage and Selection Operator (LASSO), dynamically adjusting their relative weights to enhance both coding quality and reconstruction stability. Experiments on typical shock wave datasets show that, compared with Discrete Cosine Transform (DCT), KSVD, and feature-based segmented dictionary methods (termed CC-KSVD), DH-KSVD reduces average training time by 46.4%, 31%, and 13.7%, respectively. At a Compression Ratio (CR) of 0.7, the Root Mean Square Error (RMSE) decreases by 67.1%, 65.7%, and 36.2%, while the Peak Signal-to-Noise Ratio (PSNR) increases by 35.5%, 39.8%, and 11.8%, respectively. The proposed algorithm markedly improves training efficiency and achieves lower RMSE and higher PSNR under high compression ratios, providing an effective solution for compressing long-duration, transient shock wave signals. Full article

Background: The precise role of RNA modification in post-traumatic stress disorder (PTSD) remains incompletely understood. This study aims to elucidate the effects of five common RNA modifications in PTSD, specifically m⁶A, m⁵C, m¹A, m⁷G, and ψ. Methods: We extracted data from the GEO repository to conduct a series of bioinformatics analyses. These included differential analysis to identify key regulators of five common RNA modifications, model construction using random forest (RF), least absolute shrinkage and selection operator (LASSO), and nomogram techniques, as well as consensus clustering of RNA modification subtypes. Furthermore, GO enrichment analysis was performed on DEGs associated with various RNA modification patterns. Immune cell infiltration was assessed using PCA and ssGSEA. RT-qPCR was performed to validate RNA modification-related genes (RMGs). Results: Twenty-one differentially expressed RMGs were identified. LASSO and RF intersection yielded eight signature genes (YTHDC1, IGFBP1, IGF2BP1, ALKBH5, NSUN4, TET2, TET3, WDR4) that robustly diagnosed PTSD (AUC = 0.804). Furthermore, these feature genes were validated using RT-qPCR, which was basically consistent with the results of bioinformatics analysis. Consensus clustering analysis may reveal two distinguishable subtypes: clusterA marked by high immunoinflammation, and clusterB characterized by high-neuroendocrine dysregulation. Conclusions: RMGs may play a crucial role in the pathogenesis of PTSD. Analyzing RNA modification patterns could offer potential diagnostic markers and help to guide immunotherapeutic approaches or neurotransmitter system interventions for PTSD in the future. Full article

Background/Objectives: Behavioral variant frontotemporal dementia (bvFTD), the most prevalent clinical subtype within the frontotemporal lobar degeneration spectrum disorders, is characterized by early and prominent changes that significantly disrupt everyday functioning. This study aims to identify the key correlates of functional status in bvFTD by investigating the relative contributions of cognitive deficits, behavioral disturbances, personality changes, and brain perfusion abnormalities. Additionally, it seeks to develop a theoretical framework to elucidate how these factors may interconnect and shape unique functional profiles. Methods: A total of 26 individuals diagnosed with bvFTD were recruited from the 2nd Neurology Clinic of “AHEPA” University Hospital in Thessaloniki, Greece, and underwent a comprehensive neuropsychological assessment to evaluate their cognitive functions. Behavioral disturbances, personality traits, and functional status were rated using informant-based measures. Regional cerebral blood flow was assessed using Single Photon Emission Computed Tomography (SPECT) imaging to evaluate brain perfusion patterns. Penalized Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis was performed to identify the most robust correlates of functional impairment, followed by path analyses using structural equation modeling to explore how these factors may interrelate and contribute to functional disability. Results: The severity of negative behavioral symptoms (e.g., apathy), conscientiousness levels, and performance on neuropsychological measures of semantic verbal fluency, visual attention, visuomotor speed, and global cognition were identified as the strongest correlates of performance in activities of daily living. Neuroimaging analysis revealed hypoperfusion in the right prefrontal (Brodmann area 8) and inferior parietal (Brodmann area 40) cortices as statistically significant neural correlates of functional impairment in bvFTD. Path analyses indicated that reduced brain perfusion was associated with attentional and processing speed deficits, which were further linked to more severe negative behavioral symptoms. These behavioral disturbances were subsequently correlated with declines in global cognition and conscientiousness, which were ultimately associated with poorer daily functioning. Conclusions: Hypoperfusion in key prefrontal and parietal regions, along with the subsequent cognitive and neuropsychiatric manifestations, appears to be associated with the pronounced functional limitations observed in individuals with bvFTD, even in early stages. Understanding the key determinants of the disease can inform the development of more targeted, personalized treatment strategies aimed at mitigating functional deterioration and enhancing the quality of life for affected individuals. Full article

Emerging evidence highlights the critical role of phagocytosis-related genes in CRC progression, underscoring the need for novel phagocytosis-based prognostic models to predict clinical outcomes. In this study, a four-gene (SPHK1, VSIG4, FCGR2B and FPR2) signature associated with CRC prognosis was developed using single-sample gene set enrichment analysis (ssGSEA), least absolute shrinkage and selection operator (LASSO) regression, and univariate Cox analysis. Pathway enrichment analysis was conducted on the prognostic genes, along with evaluations of the tumor microenvironment and sensitivity to immunotherapy and chemotherapy across the high- and low-risk groups. Prognostic gene validation was performed via quantitative real-time polymerase chain reaction (qRT-PCR) and immunohistochemistry (IHC) using CRC cDNA and tissue microarrays. High-risk patients showed enhanced responsiveness to immunotherapy, while chemotherapy sensitivity varied across risk subgroups. qRT-PCR results revealed upregulation of SPHK1 and FPR2 in cancer tissues, whereas FCGR2B and VSIG4 were downregulated. IHC assays confirmed increased SPHK1 and FPR2 expression in cancer samples. Single-cell RNA sequencing analysis demonstrated a decrease in SPHK1 and FCGR2B, while VSIG4 and FPR2 progressively increased during macrophage differentiation. These findings provide a potential framework for targeted therapy. Full article

Background: Cancer stem cells (CSCs) are key drivers of tumorigenesis and metastasis. However, the precise roles of CSC-associated genes in these processes remain unclear. Methods: This study integrates cancer stem cell biomarkers and clinical data from The Cancer Genome Atlas (TCGA) specific to bladder cancer (BLCA). By combining differentially expressed genes (DEGs) from TCGA-BLCA samples with CSC-related biomarkers, we conducted comprehensive functional analyses and developed an 8-gene prognostic signature through Cox regression, least absolute shrinkage and selection operator (LASSO) analysis, and multivariate Cox regression. This model was validated with GEO datasets (GSE13507 and GSE32894), and the single-cell RNA seq dataset GSE222315 was subsequently analyzed to characterize the signature genes and elucidate their interactions. And a nomogram was created to stratify TCGA-BLCA patients into risk categories. The ‘oncoPredict’ algorithm based on the GDSC2 dataset assessed drug sensitivity in BLCA. Result: From the TCGA cohort, 665 CSC-related genes were identified, with 120 showing significant differential expression. The 8-gene signature (ALDH1A1, CBX7, CSPG4, DCN, FASN, INHBB, MYC, NCAM1) demonstrated strong predictive power for overall survival in both TCGA and GEO cohorts, as confirmed by Kaplan–Meier and ROC analyses. The nomogram, integrating age, tumor stage and risk scores, demonstrated high predictive accuracy. Additionally, the oncoPredict algorithm indicated varying drug sensitivities across patient groups. Based on retrospective data, we identified a novel CSC-related prognostic signature for BLCA. This finding suggests that targeting these genes could offer promising therapeutic strategies. Full article

Radiomics has shown remarkable potential in predicting cancer prognosis by noninvasive and quantitative analysis of tumors through medical imaging. This review summarizes recent advances in the use of radiomics across various cancer types and imaging modalities, including computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, positron emission tomography (PET), and interventional radiology. Innovative sampling methods, including deep learning-based segmentation, multiregional analysis, and adaptive region of interest (ROI) methods, have contributed to improved model performance. The review examines various feature selection approaches, including least absolute shrinkage and selection operator (LASSO), minimum redundancy maximum relevance (mRMR), and ensemble methods, highlighting their roles in enhancing model robustness. The integration of radiomics with multi-omics data has further boosted predictive accuracy and enriched biological interpretability. Despite these advancements, challenges remain in terms of reproducibility, workflow standardization, clinical validation and acceptance. Future research should prioritize multicenter collaborations, methodological coordination, and clinical translation to fully unlock the prognostic potential of radiomics in oncology. Full article

Efficient power grid operations and effective business strategies require accurate prediction of power outages. However, predicting outages is a difficult task due to the large amount of heterogeneous, random, intermittent, and non-linear power grid data characterised by highly complex variable relationships. Attempting to simultaneously quantify these characteristics using a conventional single (linear or nonlinear) model may lead to inaccurate and costly results. To address this, we propose a hybrid RVM-WT-AdaBoostRT-RF framework using power grid data from the Electricity Supply Commission (Eskom) of South Africa. To achieve model interpretability, the least absolute shrinkage and selection operator (LASSO) is first applied to remedy the adverse effects of multicollinearity through regularisation and variable selection. Secondly, a random forest (RF) is used to select the top 10 most influential variables for each season for further analysis. A relevance vector machine (RVM) captures complex nonlinear relationships separately for each season, while the wavelet transform (WT) decomposes residuals generated from RVM into different frequency subseries (with reduced noise). These subseries are predicted with minimal bias using AdaBoost with regression and threshold (AdaBoostRT). Finally, we stack RVM, AdaBoostRT, RF, and residual individual predictions using RF as a meta-model to produce the final forecast with minimal error accumulation and efficiency. The comparative study, based on point forecast metrics, the Diebold-Mariano test, and prediction interval widths, shows that the proposed model outperforms vector autoregressive (VAR), RF, AdaBoostRT, RVM, and Naïve models. The study results can be utilised for optimising resource allocation, effective power grid management, and customer alerts. Full article

Background/Objective: Frailty, a geriatric syndrome characterized by decreased reserve and resistance to stressors in older adults, has been established as a robust predictor of health outcomes. Recently, the Neurocognitive Frailty Index (NFI) was introduced, including 42 physical and cognitive elements that collectively assess an individual’s vulnerability to age-related health decline. While this multidimensional approach improves predictive accuracy for cognitive decline, its high dimensionality might be a barrier to widespread adoption. Methods: We employed several machine learning techniques to reduce the dimensions of NFI while maintaining its predictive power. We trained five models: Network Analysis, neural networks, Least Absolute Shrinkage and Selection Operator Regression (LASSO), Random Forest, and eXtreme Gradient Boosting (XGBoost). Each model was calibrated using a dataset from the Canadian Study of Health and Aging, which included various cognitive, health, and functional variables. Results: Results indicated that six variables had minimal impact on outcome. This reduction in dimensionality resulted in a modified NFI scale including 36 elements, while maintaining good predictive performance for cognitive change similar to the original NFI. Conclusions: Our findings support the feasibility of applying machine learning techniques to modify predictive models in neurodegenerative diseases beyond frailty assessment. We recommend exploring the application of this scale using other data. The results also emphasize the potential of machine learning approaches for improving predictive models, highlighting their value as a tool for advancing our understanding of aging and its complexities. Full article

This study aims to develop and validate a multisequence MRI-based radiomics approach for distinguishing metaplastic breast cancer (MBC) from non-metaplastic triple-negative breast cancer (TNBC) at the initial diagnosis, which could facilitate optimal treatment selection. In this retrospective study, we analyzed 105 patients (27 MBC, 78 non-metaplastic TNBC) who underwent standardized breast magnetic resonance imaging (MRI), which included T1-weighted contrast-enhanced (T1W-CE) and short-tau inversion recovery (STIR) sequences. Two radiologists performed ground truth lesion segmentation, verified by a senior radiologist. We extracted 214 radiomic features (using PyRadiomics) and used least absolute shrinkage and selection operator (LASSO) regression for feature selection. Seven machine learning classifiers were thoroughly evaluated using five-fold cross-validation, with performance assessed through ROC analysis and accuracy metrics. The combined T1W-CE and STIR analysis demonstrated superior diagnostic performance for distinguishing MBC from non-metaplastic TNBC (AUC = 0.845; accuracy = 81%) compared with either sequence alone (T1W only AUC = 0.805; accuracy = 80%; STIR only AUC:0.768; accuracy = 77%). Multisequence MRI radiomics can reliably distinguish between MBC and TNBC at the time of initial diagnosis. This could potentially facilitate the selection of more appropriate treatments and help avoid ineffective chemotherapy for MBC patients. Full article

This study examined how the time management behavior of first-year college students predicted their academic performance. Data on 44 objective indicators of daily time management behaviors were collected from 110 first-year students via a WeChat Mini Program, through one month of consecutive daily tracking. To identify stable predictors, Least Absolute Shrinkage and Selection Operator (LASSO) regression with 5000 bootstrap resamples was conducted, and variables with high selection frequency were subsequently entered Elastic Net regression to examine explanatory relationships. Six key behavioral indicators were found to predict overall academic performance. Subject-specific models revealed varying associations: time management behaviors appeared more influential in subjects such as Physical Education and English, while their role was less evident in Mathematics. The number and nature of retained predictors also differed across disciplines. Full article

Background and Objectives: Urinary tract infections (UTIs) are a major cause of emergency department (ED) visits and hospital admissions among older adults. Although most seniors present hemodynamically stable, a sizeable fraction deteriorate during hospitalization, and no ED-specific tool exists to identify those at greatest risk. We sought to determine risk factors for in-hospital mortality in this population and to develop a predictive model. Materials and Methods: We analyzed the MIMIC-IV-ED database (2011–2019) and enrolled culture-confirmed UTI patients aged ≥ 65 years who were hemodynamically stable—defined as a systolic blood pressure ≥ 100 mm Hg without vasopressor support. Demographics, comorbidities, triage vital signs, and initial laboratory tests were extracted. Least Absolute Shrinkage and Selection Operator (LASSO) regression with 10-fold cross-validation was performed for variable selection. Discrimination was quantified with the C-statistic, calibration with the Hosmer–Lemeshow test, and clinical utility with decision curve analysis. Internal validation was assessed via 1000-sample bootstrap resampling. Results: Among 1571 eligible encounters (median age 79 years, 33% male), in-hospital mortality was 4.5%. LASSO selected eight variables; six remained significant in multivariable analysis: age, systolic blood pressure, oxygen saturation, white blood cell count, red cell distribution width, and blood urea nitrogen. The predictive nomogram demonstrated a C-statistic of 0.73 (95% CI 0.66–0.79) and outperformed traditional early warning scores. Conclusions: A six-variable nomogram may stratify mortality risk in hemodynamically stable older adults with UTI. Because the model was developed in a single U.S. tertiary-care ED, it remains hypothesis-generating until validated in external, multicenter cohorts to confirm generalizability. Full article

Background/Objectives: The prognosis of patients with hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF) is significantly affected by inflammatory state and immune dysregulation. The systemic inflammatory response index (SIRI), which reflects neutrophil, monocyte, and lymphocyte dynamics, has emerged as a potential marker of immune-inflammatory status. However, its role in predicting HBV-ACLF outcomes remains unclear. This research aims to elucidate the prognostic value of SIRI and its dynamic changes combined with disease severity scores in predicting the outcomes of HBV-ACLF. Methods: The study included HBV-ACLF patients enrolled in a multicenter clinical study between July 2019 and April 2024. Based on 90-day outcomes, the participants were categorized into survival and death groups. Clinical data and SIRI values were collected on days 0 (baseline), 3, 7, and 14. Independent prognostic factors were identified using Cox regression and least absolute shrinkage and selection operator (LASSO) analysis. The predictive value of dynamic SIRI changes combined with disease severity scores was evaluated using receiver operating characteristic (ROC) curves. Results: A total of 153 patients with HBV-ACLF were analyzed, including 104 in the survival group and 49 in the death group. SIRI values were significantly lower in the survival group than in the death group across all time points. Multivariate Cox regression analysis identified that an increased ΔSIRI at day 3 (ΔSIRI3), a higher MELD score, and a lower albumin level were independently associated with increased 90-day mortality. The combination of SIRI on day three (SIRI3) and MELD-Na score on day three (MELD-Na3) demonstrated the highest predictive performance, with an AUC of 0.817 (95% CI: 0.750–0.883). Conclusions: The combination of the SIRI and MELD-Na score on day three provides a strong predictive value for the short-term prognosis of HBV-ACLF, highlighting its potential utility in early prognostic evaluation. Full article

Influenza A virus (IAV) infection triggers excessive activation of PANoptosis—a coordinated form of programmed cell death integrating pyroptosis, apoptosis, and necroptosis—which contributes to severe immunopathology and acute lung injury. However, the molecular regulators that drive PANoptosis during IAV infection remain poorly understood. In this study, we integrated bulk and single-cell RNA sequencing (scRNA-seq) datasets to dissect the cellular heterogeneity and transcriptional dynamics of PANoptosis in the influenza-infected lung. PANoptosis-related gene activity was quantified using the AUCell, ssGSEA, and AddModuleScore algorithms. Machine learning approaches, including Support Vector Machine (SVM), Random Forest (RF), and Least Absolute Shrinkage and Selection Operator (LASSO) regression, were employed to identify key regulatory genes. scRNA-seq analysis revealed that PANoptosis activity was primarily enriched in macrophages and neutrophils. Integration of transcriptomic and computational data identified cathepsin B (CTSB) as a central regulator of PANoptosis. In vivo validation in an IAV-infected mouse model confirmed elevated expression of PANoptosis markers and upregulation of CTSB. Mechanistically, CTSB may facilitate NLRP3 inflammasome activation and promote lysosomal dysfunction-associated inflammatory cell death. These findings identify CTSB as a critical mediatoCTSBr linking lysosomal integrity to innate immune-driven lung injury and suggest that targeting CTSB could represent a promising therapeutic strategy to alleviate influenza-associated immunopathology. Full article