Search Results (5,142)

Background: Older adults with cancer are at increased risk of severe treatment-related toxicity. CARG, CRASH, and CARG-BC were developed as toxicity-risk prediction models, whereas G8 was developed as a geriatric screening instrument but has also been evaluated as a predictor of treatment-related toxicity. This study aimed to assess, separately for each instrument, the accuracy with which these tools identify older adults who develop severe chemotherapy-related toxicity. Methods: A systematic review and meta-analysis were conducted in accordance with PRISMA 2020 guidelines. Seven databases were searched through May 2026 for observational studies evaluating the predictive accuracy of the CARG, CRASH, CARG-BC and G8 tools in patients aged ≥65 years initiating chemotherapy. Pooled sensitivity and specificity with 95% confidence intervals (CIs) were calculated, and ROC curves were constructed. Risk of bias was assessed using QUADAS-2 and certainty of evidence was evaluated using GRADE. Results: Twenty-one studies were included, with an overall toxicity prevalence of 52.6%. CARG demonstrated a pooled sensitivity of 79.7% and specificity of 38.3% (AUC = 0.632). CRASH showed sensitivity of 86.9% and specificity of 68.2% (AUC = 0.866), but estimates were based on only four studies. CRASH hematological toxicity showed sensitivity of 75.9% and specificity of 53.1% (AUC = 0.694), with substantial heterogeneity. G8 yielded sensitivity of 69.5% and specificity of 41.5% (AUC = 0.666). CARG-BC showed sensitivity of 83.3% and specificity of 54.4% (AUC = 0.763), based on two breast cancer studies. Certainty of evidence ranged from low to very low. Conclusions: The instruments have distinct purposes and were not pooled against one another. CARG and G8 may be useful for initial risk screening, whereas CRASH showed a more balanced profile but remains supported by limited evidence. None should be used as a stand-alone basis to withhold or modify treatment. Full article

Osteoarthritis (OA) of the knee is characterized by cartilage matrix degeneration, which eventually leads to joint dysfunction and surgical replacement. Preoperative radiography, the standard for treatment decision making, provides limited information on the extent of degeneration, emphasizing the need for quantitative intraoperative techniques. This study evaluates the potential of vibroacoustic signals generated from instrument–tissue interactions to assess cartilage degeneration severity. A total of 136 ex vivo cartilage specimens of varying degeneration severity, histologically graded using the OARSI score, were collected from 41 patients undergoing arthroplasty. The specimens were classified into three groups: healthy cartilage (OARSI $\le 2.0$), degenerated cartilage (OARSI ≥ 2.5), and subchondral bone. Vibroacoustic signals were captured during specimen palpation using a vibration measurement system affixed to a surgical probe. 26 characteristic signal features were extracted using Continuous-Wavelet Transformation, and their discriminative power was assessed using Support Vector Machine and k-Nearest-Neighbor classifiers under patient- and specimen-level grouped cross-validation with nested hyperparameter tuning. Bone was distinguished from cartilage with 84–$91\%$ recall. Healthy cartilage (OARSI $\le 2.0$) was identifiable with 74–$80\%$ sensitivity, whereas the specificity for degenerated cartilage (OARSI ≥ 2.5) remained limited (40–$57\%$), constituting the principal limitation of the approach. Interpreting the binary cartilage classification as a diagnostic test, a receiver-operating characteristic (ROC) analysis yielded an area under the curve (AUC) of up to $0.66$ ($95\%$ CI [$0.56$, $0.76$]). The study demonstrates that vibroacoustic signals from instrument–tissue interactions may provide complementary information for intraoperative cartilage evaluation and may contribute to decision making in arthroplasty, while the reliable grading of cartilage degeneration remains an open challenge. Full article

Climate change may redistribute understory medicinal plants, but national predictions are useful when model uncertainty and field constraints are explicit. We aimed to identify the current and future suitable habitat of Sceptridium ternatum and translate predictions into survey and conservation priorities. We compiled 187 occurrences, screened 104 candidate predictors to 9, calibrated MaxEnt with kuenm, and evaluated it using partial ROC, omission rate, AICc, random and spatial-block cross-validation, and the Boyce index. Future suitability was projected with two GCMs under SSP1-2.6 and SSP5-8.5. October precipitation had the highest permutation importance, and growing-season thermal accumulation was another major contribution. NDVI had high percent contribution but no unique permutation importance. The selected model achieved mean random-fold and spatial-block test AUCs of 0.9217 and 0.8575, respectively, and a Boyce index of 0.8211. Late-century SSP5-8.5 reduced total and highly suitable habitat by 45.06% and 89.96%, respectively. Protected areas covered only 14.77% of highly suitable habitat. The resulting framework separates regional climatic suitability from habitat compatibility and management feasibility, a distinction relevant to other data-limited understory medicinal plants. It distinguishes sites for population verification, germplasm collection, and long-term monitoring, and plot-scale decisions require measurements of canopy, moisture, substrate, population state, and mycorrhizal context. Full article

Amplification of MDM2 is the molecular hallmark of atypical lipomatous tumor/well-differentiated liposarcoma (ALT/WDL) and dedifferentiated liposarcoma (DDL). While fluorescence in situ hybridization (FISH) is widely used for diagnosis, the diagnostic and prognostic value of droplet digital PCR (ddPCR) remains poorly defined. We retrospectively analyzed 341 primary adipocytic tumors, including 85 liposarcomas (22 DDL), using ddPCR to quantify MDM2 copy number variation (CNV). Diagnostic performance was compared with FISH, and associations with clinicopathological variables and outcomes were evaluated using non-parametric tests, ROC analysis, survival analysis, and Firth’s penalized Cox models. Comparison with FISH confirmed the diagnostic utility of ddPCR for detecting MDM2 amplification, while quantitative CNV assessment provided additional prognostic information. CNV values were significantly higher in deep-seated and dedifferentiated tumors and were strongly associated with local recurrence. ROC analysis identified a threshold of 16.85 copies predicting both dedifferentiation and recurrence (AUC 0.982 and 0.942, respectively). Patients with CNV ≥ 16.85 copies had significantly shorter recurrence-free and overall survival. In multivariable analysis, high CNV remained an independent predictor of recurrence (HR 36.6, 95% CI 4.0–4914.2). These findings support ddPCR as a robust method for MDM2 assessment and suggest that quantitative MDM2 copy number may improve both diagnosis and risk stratification in liposarcoma. Full article

Background/Objectives: Adverse childhood experiences (ACEs) are associated with child socio-emotional difficulties, but when using parents’ reports, it remains unclear which scoring approaches are most informative about younger children. This study evaluated a Portuguese caregiver-report ACE assessment derived from the Finkelhor/Turner framework in a community sample of children aged 2–10 years. Methods: Caregivers provided reports on ACE exposure and socio-emotional adjustment for 341 children. Three ACE scores were compared: an approximate original ACE score, a full ACE pool score, and an adapted reduced caregiver-report score. Associations with socio-emotional adjustment were examined using correlations, adjusted regressions, ROC analyses, domain models, and latent class analysis. Results: Higher ACE exposure was associated with greater difficulties and lower prosocial behavior. The nine-item approximate original ACE score, which did not include emotional neglect, showed the strongest overall predictive and discriminative performance for caregiver-reported SDQ outcomes in this Portuguese sample. This finding should therefore be interpreted as evidence regarding the available approximation of conventional ACE indicators rather than the complete original ACE inventory. Among the two expanded ACE scoring approaches, however, the adapted reduced score consistently outperformed the full ACE pool, supporting a more parsimonious approach. ROC analyses indicated acceptable discrimination for elevated total difficulties, with the approximate original score showing the highest area under the curve (AUC); however, positive predictive values were low across all three scores, indicating limited stand-alone screening utility at the individual level. Domain analyses highlighted economic stressors, maltreatment, and family disorder as salient predictors. Latent class analysis identified low adversity, moderate family/economic adversity, and high polyadversity profiles. Both adversity-exposed classes showed poorer socio-emotional adjustment than the low-adversity class. Conclusions: Findings support the relevance of caregiver-report ACE assessment for understanding socio-emotional adjustment in young children, while also highlighting the need for developmentally sensitive, culturally adapted, and outcome-validated scoring approaches. More extensive ACE pools were not necessarily more informative than shorter empirically or theoretically grounded scores. Full article

Corporate bankruptcy early warning has often been treated as a binary classification task, yet financial distress is better understood as the outcome of interacting financial conditions that must be interpreted within practical risk-management contexts. To address this issue, this study proposes a systems-oriented explainable artificial intelligence framework for corporate bankruptcy early warning. The framework implements a Hierarchical LFRM-MACI architecture in which the Local Feature Refinement Module (LFRM) refines representations within profitability, solvency, liquidity, efficiency, and growth/shareholder performance subsystems, and cross-subsystem attention models their interactions. The framework is evaluated on the public UCI Taiwanese Bankruptcy Prediction dataset under balanced and moderately imbalanced training settings. For reporting clarity, the benchmark methods are classified into traditional machine learning methods and deep learning methods; traditional tabular learners are included in the formal single-split empirical comparison, while the proposed method’s contribution is positioned as a structured deep representation with an attribution workflow rather than as an overall superiority claim over traditional machine learning models. In the single 70%/30% validation split, the proposed model obtains ROC-AUC values of 0.9286 and 0.9269 under the 1:1.0 and 1:2.5 settings, respectively. In repeated 5-fold cross-validation with five repetitions, its ROC-AUC is 0.8940 [0.8777, 0.9103] under 1:1.0 and 0.9135 [0.9001, 0.9269] under 1:2.5. To examine the interpretability of the predictions, Permutation Feature Importance (PFI) and SHAP are applied to identify subsystem-level attribution patterns across major financial subsystems. The explanation results highlight influential predictors associated with leverage pressure, profitability and asset structure, liquidity, operating efficiency, and growth/shareholder performance. These findings indicate that explainable AI can support corporate bankruptcy early warning when predictive benchmarking is combined with transparent and auditable attribution analysis for financial decision-making. Full article

Background: Anaplastic Lymphoma Kinase (ALK) is an oncogenic receptor tyrosine kinase implicated in several cancers. Despite the clinical success of ALK inhibitors, acquired resistance continues to drive the search for novel chemotypes. We developed a multiclass machine learning framework to classify ALK inhibitory activity using a curated ChEMBL dataset. Methods: Models were built using 2D molecular descriptors together with MACCS and ECFP4 fingerprints. Three widely used algorithms, Support Vector Machine (SVM), Random Forest (RF), and XGBoost, were applied for model development. Results: RF and XGBoost models demonstrated the best performance, achieving accuracies of ~0.75–0.79 with consistently high ROC–AUC values, particularly for fingerprint-based features. Bemis–Murcko scaffold analysis identified enriched chemotypes and underexplored scaffolds for further prioritization. The validated models were subsequently used to screen the Maybridge library, and compounds predicted to possess potential ALK inhibitory activity were prioritized for further computational evaluation. Applicability-domain filtering confirmed that the selected compounds occupied the predicted ALK inhibitor chemical space across multiple activity classes. The shortlisted compounds were subsequently evaluated by molecular docking to characterize their binding modes and interactions. Three candidate hits (SCR00078, SCR00073, and AW01085) were selected for further evaluation using 500 ns molecular dynamics simulations alongside the reference inhibitor Brigatinib. Simulation analyses revealed stable protein–ligand complexes and reduced conformational fluctuations relative to apo ALK, while MM/PBSA calculations identified SCR00078 and AW01085 as the most favorable binders. Conclusions: This integrated ML-to-simulation workflow prioritizes structurally novel candidate hits with predicted ALK inhibitory activity and provides an effective strategy for scaffold discovery and hit prioritization. Full article

Student dropout remains a persistent challenge in higher education, with significant academic, social and institutional implications. Although machine learning (ML) models for dropout prediction have proliferated, most select input variables based on data availability rather than theory and rarely report psychometric validation of the constructs used. This study addresses both gaps by developing and validating an early warning system at a Mexican university using exclusively pre-enrollment data from more than 46,000 student records from 2014 to 2025. Following the CRISP-DM methodology, eight theoretically grounded constructs, anchored in Tinto’s integration model, Bean’s attrition model, and Cabrera et al.’s persistence model, were operationalized from institutional intake questionnaires and assessed for internal consistency using Cronbach’s alpha prior to model training. Eight supervised ML algorithms were benchmarked across distance-based (Logistic Regression, SVM, AdaBoost, ANN) and tree-based (Random Forest, XGBoost, LightGBM, CatBoost) families. A tuned and isotonically calibrated Random Forest achieved the best overall performance (recall = 0.743, F1 = 0.524, ROC-AUC = 0.734, PR-AUC = 0.478) on a strictly held-out test set that was not used at any stage of model development. The 2023–2025 cohorts, whose dropout labels were not yet observable under the institutional definition, were scored prospectively to generate operational risk profiles. SHAP analysis identified high-school GPA, parental education, and household asset indices as dominant predictors, which mapped directly onto the three theoretical frameworks. These findings demonstrate that psychometrically grounded pre-enrollment data alone can support an operationally deployable dropout detection system, enabling proactive, evidence-based retention interventions from the first day of enrollment. Full article

(1) Background: The disaggregated and software-defined nature of fifth-generation (5G) core networks and the Open Radio Access Network (O-RAN) architecture increase the attack surface and produce large volumes of heterogeneous evidence that must be analyzed in real time to support incident reconstruction. Open-source 5G stacks (including Open5GS and srsRAN) have become reference platforms in the literature, yet recent research, such as the RANsacked study that reported 119 vulnerabilities and 97 unique CVEs across multiple LTE/5G implementations, have highlighted the pressing need for AI-based detection and forensic capabilities specific to these stacks. (2) Methods: We introduce an experimental framework consisting of a reproducible srsRAN+Open5GS testbed and an AI-driven forensic and detection pipeline. The pipeline receives control-plane (NAS, NGAP, F1AP) and Service-Based Interface (SBI) traffic, extracts protocol- and statistically grounded features and classifies traffic into seven attack types using a hybrid CNN–LSTM model. Integrity-protected and timeline-correlated forensic artifacts (PCAP, logs, memory dumps) assist in reconstructing an incident. (3) Results: The proposed hybrid model achieves a macro F1-score of 0.972 and an AUC-ROC of 0.995 (5-fold CV) and degrades gracefully under load. (4) Conclusions: We show that AI-based detection can be coupled with a scientifically sound evidence chain in open-source 5G stacks deployed as disaggregated mobile networks. Full article

Background/Objectives: Intussusception is a common pediatric emergency characterized by bowel telescoping, often leading to ischemia, necrosis, and perforation if untreated. Early identification of prognostic biomarkers could improve management strategies. This study aimed to evaluate the association between admission biomarkers and treatment outcomes in children with intussusception, to identify potential predictors. Methods: A retrospective analysis was conducted on 141 pediatric patients diagnosed with intussusception over ten years. Patients were stratified into three groups based on treatment outcome: spontaneous resolution, hydrostatic reduction, or surgery. Data collected included demographic, clinical, radiological, and laboratory parameters. Results: Among the cohort, 34% experienced spontaneous resolution, 27.7% underwent hydrostatic reduction, and 38.3% required surgery. Biomarkers such as erythrocyte count, hemoglobin, hematocrit, and platelets differed significantly across groups. Multivariate analysis identified erythrocyte count, platelet count, and potassium as variables independently associated with surgical management. Lower erythrocyte counts, higher platelet counts, and lower potassium were associated with increased likelihood of surgical intervention. The erythrocyte count, platelet count, and serum potassium showed ROC AUCs of 0.64, 0.65, and 0.70, respectively. Conclusions: Admission erythrocyte count, platelet count, and serum potassium were independently associated with the need for surgery but each showed only modest standalone discrimination. They are therefore best regarded as candidate variables for multivariable or composite predictive models rather than as standalone, clinically actionable triage markers. Further prospective studies are warranted to validate their incremental value within such models. Full article

Detecting soil–water conditions ranging from dry to fully ponded in rice fields using solely multispectral remote sensing is crucial for irrigation water management practices focused on water savings in Semi-Arid Australia. To this end, this research employed the Minimum Redundancy Maximum Relevance (mRMR) algorithm to identify a set of multispectral remote sensing indices for use with Machine Learning (ML) to predict three soil–water conditions in irrigated rice: “Flooded”, “Saturated”, and “Dry”. Two models were developed: Model 1, using the most frequently used remote sensing indices in the literature; Model 2, including multispectral variables selected by the mRMR algorithm. Model 2 achieved the highest performance, with an accuracy of 0.64 and a $\mathrm{k}\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{a}$ of 0.46, and ROC-AUC values of 0.87, 0.62, and 0.83 for “Flooded”, “Saturated”, and “Dry”, respectively. All models exhibit high confusion rates between “Flooded” and “Saturated” conditions, suggesting that multispectral remote sensing doesn’t provide sufficient information to distinguish these soil–water conditions. The SHAP analysis revealed that vegetation-sensitive indices encoding information on crop biomass, plant moisture, and senescence status were the primary drivers of soil–water condition prediction. Full article

Background: Conventional anthropometric indices such as body mass index (BMI), waist circumference (WC), and waist-to-height ratio (WHtR) are widely used for obesity screening; however, their accuracy may be limited in young Japanese women, who often exhibit higher percentage body fat (%BF) at lower BMI values. This pilot study examined the relationships between DXA-derived %BF and anthropometric indices and evaluated their ability to identify excess adiposity. Methods: A cross-sectional study was conducted among 78 Japanese female university students aged 18–28 years. Whole-body composition was assessed using dual-energy X-ray absorptiometry (DXA), and excess adiposity was defined as %BF ≥ 30%. Receiver operating characteristic analyses were performed to evaluate the discriminatory ability of anthropometric indices and identify exploratory optimal cut-off values. Results: Participants had a median BMI of 20.3 kg/m² and a mean %BF of 32.3 ± 5.4%. Conventional BMI, WC, and WHtR thresholds showed very low sensitivity for detecting excess adiposity despite perfect specificity. ROC analyses identified substantially lower optimal cut-off values, yielding sensitivities of approximately 70–80%, specificities of 58–75%, and AUC values of 0.76–0.78. Among novel anthropometric indices, body roundness index (BRI) demonstrated the strongest discriminatory ability. Conclusions: Conventional anthropometric thresholds may underestimate DXA-defined excess adiposity in young Japanese female university students. This exploratory pilot study identified lower candidate cut-off values and suggested that BRI may serve as a useful adjunctive anthropometric index for identifying excess adiposity in this population. These findings should be interpreted cautiously and require validation in larger and more diverse populations before clinical application. Full article

Flooding is one of the most damaging natural hazards worldwide, particularly in data-scarce watersheds where long-term hydrometeorological records are limited. This study focuses on the R’Dom watershed in the Meknes region, Morocco, and aims to improve flood susceptibility and relative flood risk assessment by integrating remote sensing, Geographic Information Systems (GIS), and multi-criteria decision-making (MCDM) approaches. The Analytic Hierarchy Process (AHP) and Fuzzy Analytic Hierarchy Process (Fuzzy AHP/FAHP) were applied to evaluate flood hazard, vulnerability, and overall risk using seventeen conditioning factors, including topographic, hydrological, geological, land-cover, socio-economic, and infrastructure-related variables. The Flood Hazard Index (FHI), Flood Vulnerability Index (FVI), and Flood Risk Index (FRI) were calculated to produce flood susceptibility, vulnerability, and relative flood risk maps. Model validation was performed using a point-based flood inventory dataset composed of 900 locations, including 450 flood and 450 non-flood points, compiled from historical flood information, field observations, local information, official reports, and satellite-based interpretation. The dataset was divided into 70% for training and 30% for testing, and model performance was assessed using receiver operating characteristic–area under the curve (ROC-AUC) analysis. The AHP and Fuzzy AHP models showed good to excellent predictive performance, with testing AUC values ranging from 0.767 to 0.935. The AHP-based models achieved the highest testing performance, while Fuzzy AHP remained useful for representing uncertainty in expert judgment and gradual spatial transitions. The final flood risk map indicates that approximately 18.78% of the study area, corresponding to 240.62 km², is classified as having a high to very high flood risk, mainly around the Meknes conurbation and locally near the El Hajeb region. These results provide a practical decision-support tool for identifying priority areas for flood mitigation, land-use planning, and watershed management. However, the proposed GIS–MCDA approach produces relative flood susceptibility and risk classes and does not replace hydrological or hydraulic modeling for estimating flood depth, discharge, inundation extent, or return-period-based flood hazard. Full article

Falls among older adults can cause serious injury and loss of independence. cStick 2.0 is a vision-enabled, IoMT-edge based smart walking-stick prototype that combines multimodal fall-risk classification with embedded obstacle awareness. The fall-risk classifiers were evaluated using a 9670-record development dataset, on which the compact DNN achieved 95.40% accuracy, 92.35% balanced accuracy, a macro F1-score of 93.90%, and a ROC-AUC of 97.44%. The Arduino Nicla Vision obstacle module used an INT8 Edge Impulse model with centroid-based direction assignment and time-of-flight distance sensing; 144 controlled trials produced 75.00% obstacle-presence accuracy at approximately 19–20 FPS. Sensor acquisition, GPS, display output, buzzer response, and CSV record accumulation were demonstrated at a prototype level. Synchronized older-adult evaluation, device-to-application communication, secure caregiver services, multimodal accessibility feedback, and longitudinal personalization remain future validation stages. Full article

Natural fractures strongly influence fluid flow, hydraulic-fracturing performance, and production heterogeneity in tight oil reservoirs. Their identification from conventional logs remains challenging because image-log and core coverage is limited, fracture-related logging responses are non-unique, and discrete fracture interpretations are difficult to align with regularly sampled logging sequences. This study used conventional logging data and electrical image-log interpretations from 17 wells in the Xifeng Oilfield, Ordos Basin, together with core observations from selected intervals, to develop a fracture response mapping and triangular topology-optimized bidirectional long short-term memory model (FRM-BiLSTM-TTAO). After sliding-window construction and density-based undersampling, 1713 samples were retained and partitioned at the well level into 14 training wells and three independent test wells, yielding an approximate training-to-test sample ratio of 75:25. FRM extracts lithologic-background, local-abrupt-change, multiscale-fluctuation, and integrated fracture response features; BiLSTM captures bidirectional depth dependencies; and TTAO selects fracture response features and optimizes the network architecture and training parameters. On the test set, the model achieved a ROC-AUC of 0.9079, a recall of 0.8671, and an F1-score of 0.8464, outperforming CNN, MLP, ResNet1D, XGBoost, and the corresponding ablation models. The predicted high-probability intervals were generally consistent with image-log interpretations and core observations, indicating the feasibility of the proposed method for identifying fracture-prone intervals within the study area. Full article