Search Results (1,078)

Background/Objectives: Hepatocellular carcinoma (HCC) presents limited therapeutic options for advanced disease, and sorafenib therapy is hampered by significant interpatient heterogeneity in response. This necessitates biomarker-guided strategies to personalize treatment. This study investigated the long noncoding RNAs UCA1 and MALAT1 as pharmacogenomic biomarkers for personalizing sorafenib therapy in advanced HCC. Methods: In a prospective cohort of 154 HCC patients receiving first-line sorafenib (400 mg twice daily), serum lncRNA levels were quantified by RT-qPCR at baseline, Week 4, and Week 12. Expression levels were correlated with treatment response (mRECIST), time-to-progression (TTP), and overall survival (OS). Statistical analyses included Kaplan–Meier estimates, Cox proportional hazards models, and ROC curve analysis. Results: High baseline expression of UCA1 (77.9% of patients) and MALAT1 (73.4%) was associated with aggressive disease. High UCA1 correlated with reduced 12-month survival (60.8% vs. 73.5%, p = 0.026) and shorter median Time-to-Progression (TTP) (18.0 vs. 21.9 weeks, p = 0.002). High MALAT1 was associated with significantly shorter median TTP (18.0 vs. 25.2 weeks, p = 0.003). In multivariable analysis, both lncRNAs were independent prognostic factors for shorter TTP (UCA1: HR = 1.52, p = 0.014; MALAT1: HR = 1.61, p = 0.006). Serial monitoring revealed that a ≥10% rise in either lncRNA by Week 4 predicted a five-fold higher progression risk by Week 12 (52% vs. 10%, p < 0.001), providing a median lead time of 7.0 weeks before radiological confirmation of progression. Conclusions: These findings demonstrate that UCA1 and MALAT1 enable early identification of sorafenib resistance. Baseline stratification and serial monitoring can provide early detection of treatment resistance, informing clinical decision-making and supporting their potential utility for personalizing therapy in advanced HCC. Full article

Background/Objectives: Colorectal cancer (CRC) is a leading cause of cancer deaths worldwide, underscoring the need for diagnostic tools that early, accurate, and clinically interpretable. Current artificial intelligence (AI) models are predominantly unimodal and lack sufficient interpretability, which restricts their clinical adoption. Methods: We propose IDF-Net, an interpretable dynamic fusion framework that integrates endoscopy, computed tomography (CT), and histopathology using modality-specific encoders, a dual-stage adaptive gating mechanism, and cross-modal attention. We conducted stratified 5-fold cross-validation and assessed interpretability using spatial heatmaps and modality attribution. We also quantified the results using the intersection-over-union metric for saliency alignment. Results: IDF-Net achieved a state-of-the-art accuracy of 0.920 (0.907–0.936) and area under the curve (AUC) of 0.991 (95% CI: 0.965–0.997), significantly outperforming unimodal and static-fusion baselines (p < 0.05). Interpretability analysis of IDF-Net demonstrated a strong alignment between Gradient-weighted Class Activation Mapping++ heatmaps and expert-annotated lesions, as well as case-specific modality contributions via SHapley Additive exPlanations values. Ablation studies confirmed the contribution of each component, with dynamic routing and cross-attention fusion improving AUC by 0.038 and 0.046, respectively. Conclusions: IDF-Net introduces a dynamically fused, multimodal diagnostic framework with integrated quantitative interpretability, demonstrating superior accuracy and strong potential for clinical translation in CRC diagnosis. The model’s adaptive design allows it to function robustly even when CT data is unavailable, aligning with common clinical pathways while leveraging additional imaging when present for comprehensive staging. Full article

Background/Objectives: Post-thrombotic syndrome (PTS) following unprovoked deep vein thrombosis (DVT) lacks readily available, calibrated risk estimates at defined follow-up horizons. Building on signals that thrombus burden, care processes, and a form of metabolic–inflammatory tone influence outcomes, we prospectively evaluated survival machine-learning models, explicitly including hyperuricemia while excluding what we consider major inflammatory confounders. Methods: Adults with first-episode unprovoked lower-extremity DVT were enrolled at two centers (July 2024–September 2025). PTS (Villalta) was assessed at 3, 6, 9, and 12 months. The cohort was split 70/30 into training and test sets. Eight learners (RSF, GBM, LASSO + Cox, CoxBoost, survivalsvm, XGBoost-Cox, superpc, and plsRcox) were tuned using 10-fold cross-validation in training and once evaluated in the independent test set. Performance metrics included all time-dependent AUCs, fixed-time ROC AUCs with bootstrap 95% CIs, C-index, various forms of calibration, decision-curve analysis, and simple Kaplan–Meier risk group separation. Results: 193 patients were analyzed (PTS in 64%). High 9-month AUCs were seen in training: GBM (0.992) and RSF (0.982) being the strongest; by 12 months, both remained near constant. Test set performance followed a similar pattern, with RSF again favored (AUC 0.948) and XGBoost/GBM close behind. Calibration was satisfactory, net benefit from decision curves positive, and to a large extent, risk groups were separated as expected. Conclusions: Survival machine-learning models, at least in this dual-center prospective cohort, produced a clinically useful risk of PTS. Hyperuricemia, or any metabolically based signal, is a valuable addition to the “anatomy and care” of DVT. External validation is still required. Full article

Hydrogen energy is vital for a clean, low-carbon society, and proton exchange membrane fuel cells (PEMFCs) represent a core technology for the conversion of hydrogen chemical energy into electrical energy. When PEMFC single cells are stacked under assembly force for high power output, their porous electrodes (gas diffusion layers, GDLs; catalyst layers, CLs) undergo compressive deformation, altering internal transport processes and affecting cell performance. However, existing microscale studies on PEMFC porous electrodes insufficiently consider compression (especially in CLs) and have limitations in obtaining compressed microstructures. This study proposes a combined framework from a microstructure perspective. It integrates the finite element method (FEM) with computational fluid dynamics (CFD). It reconstructs microstructures of GDL, CL, and GDL-bipolar plate (BP) interface. FEM simulates elastic compressive deformation, and CFD calculates transport properties (solid zone: heat/charge conduction via Laplace equation; fluid zone: gas diffusion/liquid permeation via Fick’s/Darcy’s law). Validation shows simulated stress–strain curves and transport coefficients match experimental data. Under 2.5 MPa, GDL’s gas diffusivity drops 16.5%, permeability 58.8%, while conductivity rises 2.9-fold; CL compaction increases gas resistance but facilitates electron/proton conduction. This framework effectively investigates compression-induced transport property changes in PEMFC porous electrodes. Full article

The conservation and breeding of the western honey bee (Apis mellifera) is central dependent on accurate subspecies assignment, but the most commonly used methods are labor-intensive classical morphometrics and costly molecular assays. We developed an XGBoost-based classification framework using a compact set of routinely measurable characters. A curated dataset of labeled workers was measured under harmonized protocols; features were screened according to embedded importance, and model performance was assessed using five-fold cross-validation, outperforming standard machine learning baselines. The resulting model using only the top 10 characters—primarily forewing venation angles and abdominal plate metrics—achieved high performance (accuracy = 0.98; F1 = 0.99) and an area under the receiver operating characteristic curve (AUC) of 0.99 (95% CI = 0.995–0.999). SHAP analyses confirmed the discriminatory contributions of these features, while error inspection suggested that misclassifications were concentrated in morphologically overlapping lineages. The model’s performance supports its use as a rapid triage tool alongside genetic testing, providing a scalable and interpretable tool for researchers to create and deploy custom morphometric models, demonstrated here for A. mellifera but portable to other insect taxa. Full article

Background/Objectives: Intestinal ultrasound (IUS) is increasingly used to monitor ulcerative colitis (UC), but its predictive value remains unclear. This systematic review evaluated the ability of IUS parameters and scores to predict short- and long-term treatment response, remission, and adverse outcomes in hospitalized and outpatient UC populations. Methods: A systematic review was conducted according to Cochrane and PRISMA guidelines. MEDLINE and Embase were searched for prospective studies assessing IUS as a predictor of clinical or endoscopic response, remission, relapse, or adverse outcomes in adult UC. Two reviewers independently performed screening, data extraction, and QUADAS-2 assessment. Results: Eighteen prospective studies were included: eleven outpatient studies and seven involving hospitalized patients treated with intravenous corticosteroids (IVCS). In hospitalized patients, bowel wall thickness (BWT) was the most consistent predictor of treatment failure, rescue therapy, colectomy, and clinical response. Baseline BWT showed variable performance, but once IVCS was initiated, early BWT change within 48–72 h was the strongest marker of disease trajectory. Non-responders had higher BWT and smaller reductions. A BWT ≥ 4 mm, absolute reduction ≤ 1 mm, or relative reduction ≤ 20% at 48 h reliably identified patients needing rescue therapy (area under the curve (AUC) values of 0.77 (95% confidence interval (CI) 0.71–0.74), 0.71 (95% CI 0.56–0.86), and 0.74 (95% CI 0.60–0.88)). Colectomy risk was similarly predicted: BWT < 3 mm at 48 h was associated with no colectomies, whereas BWT ≥ 4 mm or persistently elevated BWT at day 6 markedly increased risk (Odds ratio (OR) 9.5-fold (95% CI 1.4–64.0) and OR 8.3 (95% CI 1.7–40.0), respectively). Other sonographic features (loss of haustration, increased vascularity) added supplementary but less consistent value. In outpatients, BWT also demonstrated the strongest predictive accuracy. BWT ≤ 3.6 mm at 2 weeks and <3.0 mm at 6 weeks were associated with early endoscopic remission (area under the receiver operating characteristic (AUROC) of 0.87 (95% CI 0.71–1.00) and 0.82 (95% CI 0.63–1.00), respectively). Dynamic changes with ≥23–25% relative reduction predicted clinical or endoscopic response (AUROC of 0.81 (95% CI 0.61–1.00) and OR of 13.9 (95% CI 1.13–1986.85), respectively). Persistent BWT > 3.5 mm or minimal reduction (<20% or <1 mm) indicated a low likelihood of long-term remission. Composite vascularity-based indices, particularly the Milan Ultrasound Criteria (MUC), strengthened prediction: MUC ≤ 4.3 or ≥2-point reduction at 12 weeks predicted long-term remission (AUROC 0.88 (95% CI 0.750–0.952) and 0.82 (95% CI 0.68–0.91), respectively), while MUC ≥ 7.7 indicated high risk of treatment failure or colectomy (AUROC 0.77 (95% CI: 0.73–0.82)). Conclusions: Across clinical settings, BWT consistently emerged as the strongest IUS predictor of UC treatment outcomes. Early BWT change within 48–72 h in hospitalized patients and absolute BWT values at 2–6 weeks in outpatients showed high predictive accuracy for response, remission, and colectomy. Composite indices incorporating vascularity further improved prediction. These findings support the incorporation of IUS into early treatment-response algorithms and underscore the need for standardized cut-offs and multicenter validation. Full article

Rosuvastatin calcium is a promising lipid-lowering agent and the drug of choice in hyperlipidemia. Conventional solid oral delivery of rosuvastatin is limited by its poor solubility and ultimately poor bioavailability. An attempt was made to fabricate the cocrystals of RSC for enhancing solubility and bioavailability. Cocrystals were prepared by a microwave synthesiser-assisted solvent evaporation technique with multiple cocrystal formers. Rosuvastatin-Ascorbic acid (RSC-AA) cocrystals showed the highest solubility (~5-fold increased) amongst all twenty drug-coformer combination (DCC). RSC-AA cocrystals (1:1 ratio) were further characterized by various analytical techniques like FTIR, DSC and XRD to confirm the formation of cocrystals. RSC-AA cocrystals also showed improved flow properties and compressibility in comparison with pure drug, and it was demonstrated using the SeDeM diagram. RSC-AA cocrystals were further formulated into an immediate-release tablet by implementing experimental optimization. Comparative dissolution study of the cocrystal and pure drug tablet revealed improved dissolution after cocrystallization. RSC-AA cocrystal tablet showed the % drug release of 95.61 ± 3.94 while RSC pure drug showed the drug release of 67.83 ± 3.29. In vivo pharmacokinetic analysis showed significant improvement in systemic availability and cumulative absorption of the drug. The peak plasma concentration (Cmax) for RSC pure drug was 13.924 ± 0.477 μg/mL, while RSC-AA cocrystals showed a peak plasma concentration of 22.464 ± 0.484 μg/mL. Area Under Curve (AUC) of RSC-AA cocrystal was also significantly greater compared to the pure drug. In the stability study analysis, the shelf life was calculated from a graphical method and was found to be around 34.58 months for RSC-AA cocrystal tablets and 19.87 months for RSC pure drug tablets, which indicates improved stability with cocrystallization. Overall, the cocrystallization resulted in significant improvement in dissolution and solubility of RSC. Full article

Background: To meet the urgent need for novel antibacterial agents that are active also against worrying superbugs, natural pentacyclic triterpenoids, including totally inactive betulin (BET) and betulinic acid (BA), as well as ursolic acid (UA), active on Gram-positive bacteria, have been chemically modified, achieving compounds 1–7. Methods: Triterpenoid derivatives 1–7 and all synthetic intermediates were characterized by chemometric-assisted FTIR and NMR spectroscopy, as well as by other analytical techniques, which confirmed their structure and high purity. Minimum inhibitory concentration values (MICs) of 1–7, BET, BA and UA were determined by the broth dilution method, using a selection of Gram-positive and Gram-negative clinically isolated superbugs. Results: Performed experiments evidenced that compounds 4–7 had potent antibacterial effects against Gram-positive methicillin-resistant Staphylococcus aureus and S. epidermidis (MRSA and MRSE), as well as against vancomycin-resistant Enterococcus faecalis and E. faecium (VRE). The antibacterial effects of 4–7 were due to the insertion of a triphenyl phosphonium (TPP) group and were higher than those reported so far for other BET, BA and UA derivatives, especially considering the complex pattern of resistance of the isolates used here and their clinical source. Conclusions: For the first time, by inserting TPP, a real activity (MICs 2–16 µg/mL) was conferred to inactive BET and BA (MICs > 1024 and 256 µg/mL). Moreover, the antibacterial effects of UA were improved 16- and 32-fold against MRSE and MRSA (MICs = 2 vs. 32 and 64 μg/mL). Future Perspectives: Based on these very promising microbiologic results, new experiments are currently underway with the best-performing compounds 5 and 7 (MICs = 2 μg/mL) on an enlarged number of Gram-positive isolates, to confirm their MICs. Moreover, investigations about their possible antibiofilm activity, time-killing curves and cytotoxicity on eukaryotic cells will be carried out to define their pharmacological behavior and clinical potential. Full article

Alcohols and esters are core flavor-active constituents of Qingxiangxing Baijiu (QXB), yet ethanol concentration’s regulatory role in their thresholds and interactions remains unclear. Physicochemical analysis showed reduced-alcohol QXB (L-QX, 42%, v/v) had higher total acid (1.48 g/L) but lower total ester (1.52 g/L) than high-alcohol QXB (H-QX, 53%, v/v; 1.20 g/L total acid, 2.05 g/L total ester). Sensory evaluation (0–5 scale) revealed H-QX had higher fruity (3.6 vs. 2.0), grassy (3.2 vs. 1.8), and grainy (3.0 vs. 1.9) aroma scores, while L-QX showed higher sour (2.1 vs. 1.5) and lees (1.7 vs. 1.1) notes (p < 0.05). The quantification of gas chromatography-flame ionization detection (GC-FID) determined the concentrations of eight alcohols and esters in H-QX samples and identified that most flavor compounds had higher concentrations than L-QX samples. Three alternative forced-choice tests showed 53% ethanol elevated olfactory thresholds (OTs) of five compounds, with ethyl lactate (1.53-fold) and isopentanol (1.89-fold) vs. 42%. For 16 alcohol–ester binary mixtures, 12 pairs had OT ratios (53% vs. 42%) < 1, especially 3 pairs (e.g., n-propanol-ethyl acetate) < 0.5. OAV/S curve analyses indicated all 16 mixtures had masking effects, with 11 pairs stronger at 42%. Verification validated 53% ethanol mitigated masking, enhancing fruity/grassy aromas by 38.1%/25.0%. This study provides support for QXB dealcoholization flavor regulation. Full article

Objective: To establish a reliable machine-learning-based model for predicting the risk of lower limb amputation in patients with diabetic foot ulcers and to provide quantitative evidence for clinical decision-making and individualized prevention strategies. Methods: This retrospective study analyzed data from 149 hospitalized diabetic foot ulcer patients treated at Beijing Shijitan Hospital between January 2019 and December 2022. Patients were divided into amputation and non-amputation groups according to clinical outcomes. Candidate predictors—including infection biomarkers, vascular parameters, and nutritional indices—were first screened using the least absolute shrinkage and selection operator algorithm. Subsequently, a support vector machine model was trained and internally validated via five-fold cross-validation to estimate amputation risk. Model performance was evaluated by discrimination, calibration, and clinical utility analysis. Results: Among all enrolled variables, C-reactive protein and Wagner grade were identified as independent predictors of amputation (p < 0.05). The optimized support vector machine model achieved excellent discrimination, with an area under the Receiver Operating Characteristic curve of 0.89, and demonstrated a moderate level of calibration (Hosmer–Lemeshow χ² = 19.614, p = 0.012). Decision curve analysis showed a net clinical benefit of 0.351 when the threshold probability was set at 0.30. The model correctly classified 82.4% of cases in internal validation, confirming its predictive robustness and potential for clinical application. Conclusions: C-reactive protein and Wagner grade are key determinants of amputation risk in diabetic foot ulcer patients. The support vector machine-based prediction model exhibits strong accuracy, clinical interpretability, and personalized interventions. Full article

Objectives: Previously, we identified four open-angle glaucoma (OAG)-associated autoantibodies (anti-ETNK1, anti-VMAC, anti-NEXN, and anti-SUN1) using proteome-wide autoantibody screening by wet protein arrays. The objective of this exploratory study was to evaluate the diagnostic performance of these four glaucoma-associated autoantibodies using automated machine learning. Methods: Plasma samples from 119 patients with OAG and 35 patients with cataracts as controls were enrolled for the study. All machine-learning analyses were performed in Python 3.9.16 (GCC 11.2.0) using scikit-learn 1.2.2 and PyCaret 3.0.1. Variables included plasma levels of the autoantibodies, age, sex, and intra-ocular pressure (IOP). Probability calibration (Platt/sigmoid and isotonic) was assessed with reliability curves and Brier scores. Model explainability was examined with permutation importance, SHAP values, and an ablation analysis removing one autoantibody at a time. Results: The tuned random forest achieved an out-of-fold (OOF) area under the receiver-operating characteristic curve (ROC–AUC) of 0.852 (±0.040), an average precision (AP) of 0.950, and an F1 score of 0.865. Isotonic mapping improved agreement between predicted and empirical probabilities. Among these four autoantibodies, VMAC was the most important factor for the model’s prediction. Conclusions: A machine learning model using four autoantibodies from blood samples showed potential for diagnosing OAG. Full article

Unmanned Aerial Vehicles (UAVs) are integral components of future 6G networks, offering rapid deployment, enhanced line-of-sight communication, and flexible coverage extension. However, UAV communications in low-altitude environments face significant challenges, including rapid link variations due to attitude instability, severe signal blockage by urban obstacles, and critical sensitivity to transmitter–receiver alignment. While traditional planar reconfigurable intelligent surfaces (RIS) show promise for mitigating these issues, they exhibit inherent limitations such as angular sensitivity and beam squint in wideband scenarios, compromising reliability in dynamic UAV scenarios. To address these shortcomings, this paper proposes and evaluates a spherical-cap reflective intelligent surface (ScRIS) specifically designed for dynamic low-altitude communications. The intrinsic curvature of the ScRIS enables omnidirectional reflection capabilities, significantly reducing sensitivity to UAV attitude variations. A rigorous analytical model founded on Generalized Sheet Transition Conditions (GSTCs) is developed to characterize the electromagnetic scattering of the curved metasurface. Three distinct 1-bit RIS unit cell coding arrangements, namely alternate, chessboard, and random, are investigated via numerical simulations utilizing CST Microwave Studio and experimental validation within a mechanically stirred reverberation chamber. Our results demonstrate that all tested ScRIS coding patterns markedly enhance electromagnetic field uniformity within the chamber and reduce the lowest usable frequency (LUF) by approximately 20% compared to a conventional metallic spherical reflector. Notably, the random coding pattern maximizes phase entropy, achieves the most uniform scattering characteristics and substantially reduces spatial field autocorrelation. Furthermore, the combined curvature and coding functionality of the ScRIS facilitates simultaneous directional focusing and diffuse scattering, thereby improving multipath diversity and spatial coverage uniformity. This effectively mitigates communication blind spots commonly encountered in UAV applications, providing a resilient link environment despite UAV orientation changes. To validate these findings in a practical context, we conduct link-level simulations based on a reproducible system model at 3.5 GHz, utilizing electromagnetic scale invariance to bridge the fundamental scattering properties observed in the RC to the application band. The results confirm that the ScRIS architecture can enhance link throughput by nearly five-fold at a 10 km range compared to a baseline scenario without RIS. We also propose a practical deployment strategy for urban blind-spot compensation, discuss hybrid planar-curved architectures, and conduct an in-depth analysis of a DRL-based adaptive control framework with explicit convergence and complexity analysis. Our findings validate the significant potential of ScRIS as a passive, energy-efficient solution for enhancing communication stability and coverage in multi-band 6G networks. Full article

Background/Objectives: Cirrhosis significantly alters physiological function and drug metabolism, particularly for medications primarily metabolized by CYP2C19 and CYP3A4. This study aims to establish a physiologically based pharmacokinetic (PBPK) modelling framework capable of predicting pharmacokinetic changes across different stages of cirrhosis, and to determine optimal dosing regimens that achieve drug exposure levels comparable to those in healthy individuals. Methods: We constructed a physiologically based pharmacokinetic (PBPK) model that incorporates six drugs, including omeprazole, lansoprazole, midazolam, ondansetron, verapamil, and alfentanil, which are metabolized primarily by CYP2C19 or CYP3A4. The pharmacokinetics of these drugs following oral or injectable administration were simulated in 1000 virtual healthy subjects, and the PBPK model was validated using clinical data. The model was further adapted to account for physiological changes in cirrhotic patients, extending its application to a population of 1000 virtual patients with liver cirrhosis. Results: Most observed data fell within the 5th and 95th percentiles of the virtual patient simulation results. Additionally, for most simulations, the area under the concentration-time curve (AUC) and peak concentration (C_max) were within 0.5- to 2-fold of the observed values. Sensitivity analysis indicated that the reduced expression of metabolizing enzymes increased plasma concentrations of drugs, which was a major factor contributing to the elevated drug exposure in patients with cirrhosis. The clinical dosing regimens of the CYP2C19 substrate omeprazole and the CYP3A4 substrate ondansetron were optimized for use in cirrhotic patients. Conclusions: The developed PBPK model successfully predicted the pharmacokinetics of CYP2C19 and CYP3A4 substrates in both healthy individuals and cirrhotic patients and can be effectively used for dose optimization in cirrhotic populations. Full article

Background/Objectives: Metabolic syndrome (MetS) increases the risk of type 2 diabetes and cardiovascular disease. We aimed to identify the key metabolic predictors of MetS in a Central European cohort and to compare classical statistics with modern machine learning (ML) models. Methods: We analysed 956 adults from the Lower Silesia Healthy Donors cohort. Clinical, anthropometric, biochemical, and lifestyle variables were collected using standardised procedures. Group differences were tested with Mann–Whitney U tests and effect sizes. A multivariable logistic regression (outcome: binary MetS defined as ≥3 harmonised components, MetS_bin) estimated adjusted odds ratios. In parallel, ML models (logistic regression, Random Forest, XGBoost, LightGBM, CatBoost) were trained with stratified 5-fold cross-validation. Performance was evaluated by accuracy, F1-macro, and area under the receiver-operating characteristic curve (ROC AUC). Model interpretability used SHAP values. Results: Overweight/obese participants had higher fasting glucose (median 92.0 vs. 84.6 mg/dL), fasting insulin (9.9 vs. 6.6 µU/mL), and systolic blood pressure (134 vs. 121 mmHg) and lower HDL cholesterol (53 vs. 66 mg/dL) compared to normal-BMI individuals (all p < 0.001, r ≈ 0.39–0.41). Participants with a higher waist circumference also showed markedly increased HOMA-IR (2.16 vs. 1.34; p < 0.001). In multivariable logistic regression, waist circumference, BMI, triglycerides, HDL cholesterol, fasting glucose, and systolic blood pressure were independently associated with MetS, yielding a test ROC-AUC of 0.98 and PR-AUC of 0.88. Machine learning models further improved discrimination: Random Forest, XGBoost, LightGBM, and CatBoost all achieved very high performance (test ROC-AUC ≥ 0.99, PR-AUC ≥ 0.98), with CatBoost showing the best cross-validated PR-AUC (~0.99) and favourable calibration. SHAP analyses consistently highlighted fasting glucose, triglycerides, HDL cholesterol, waist circumference, and systolic blood pressure as the most influential predictors. Conclusions: Combining classical regression with modern gradient-boosting models substantially improves the identification of individuals at risk of MetS. CatBoost, XGBoost, and LightGBM delivered near-perfect discrimination in this Central European cohort while remaining explainable with SHAP. This framework supports clinically meaningful risk stratification—including a “subclinical” probability zone—and may inform targeted prevention strategies rather than purely reactive treatment. Full article

Three-dimensional cell model systems such as tumour organoids allow for in vitro modelling of self-organized tissue with functional and histologic similarity to in vivo tissue. However, there is a need for standard protocols and techniques to confirm the presence of cancer within organoids derived from tumour tissue. The aim of this study was to assess the utility of a Nanostring gene expression-based machine learning classifier to determine the presence of cancer or normal organoids in cultures developed from both benign and cancerous stomach biopsies. A prospective cohort of normal and cancer stomach biopsies were collected from 2019 to 2022. Tissue specimens were processed for formalin-fixed paraffin-embedding (FFPE) and a subset of specimens were established in organoid cultures. Specimens were labelled as normal or cancer according to analysis of the FFPE tissue by two pathologists. The gene expression in FFPE and organoid tissue was measured using a 107 gene Nanostring codeset and normalized using the Removal of Unwanted Variation III algorithm. Our machine learning model was developed using five-fold nested cross-validation to classify normal or cancer gastric tissue from publicly available Asian Cancer Research Group (ACRG) gene expression data. The models were externally validated using the Cancer Genome Atlas (TCGA), as well as our own FFPE and organoid gene expression data. A total of 60 samples were collected, including 38 cancer FFPE specimens, 5 normal FFPE specimens, 12 cancer organoids, and 5 normal organoids. The optimal model design used a Least Absolute Shrinkage and Selection Operator model for feature selection and an ElasticNet model for classification, yielding area under the curve (AUC) values of 0.99 [95% CI: 0.99–1], 0.90 [95% CI: 0.87–0.93], and 0.79 [95% CI: 0.74–0.84] for ACRG (internal test), FFPE, and organoid (external test) data, respectively. The performance of our final model on external data achieved AUC values of 0.99 [95% CI: 0.98–1], 0.94 [95% CI: 0.86–1], and 0.85 [95% CI: 0.63–1] for TCGA, FFPE, and organoid specimens, respectively. Using a public database to create a machine learning model in combination with a Nanostring gene expression assay allows us to allocate organoids and their paired whole tissue samples. This platform yielded reasonable accuracy for FFPE and organoid specimens, with the former being more accurate. This study re-affirms that although organoids are a high-fidelity model, there are still limitations in validating the recapitulation of cancer in vitro. Full article