Search Results (45,901)

This study evaluated the potential of Nanche seeds as a bio-adsorbent for removing TA from aqueous solutions. Additionally, we inferred the interaction mechanisms involved in the adsorption process from physicochemical evidence, isotherm modelling, and ADI analysis. The Langmuir model best fit the experimental data (R² = 0.999) across temperatures (20–60 °C). Thermodynamically, the process was endothermic and spontaneous, (∆H° = 10.11 ± 0.45 kJ/mol and ∆G° = −28.58 ± 1.03 kJ/mol), indicating that the interaction mechanisms were physical. The value of ∆S° (0.116 ± 0.003 kJ/mol·K), the Freundlich constant (n > 1), and data derived from ADI analysis of SEM micrographs that support the increased complexity and multifractal nature of the process. FTIR revealed surface changes in the adsorbent material attributable to the interactions between the seed and the TA. Extent of vibrations associated with carboxyl groups decreased after adsorption, which was attributed to low-energy interactions, such as π-π stacking, supporting physisorption interaction. The ki values increased linearly with temperature and TA concentration. This study used ideal adsorption systems to elucidate fundamental mechanisms, providing a foundation for future, more complex research. Nance seeds proved to be effective yet underutilized adsorbent material. Further studies are needed on its applicability in continuous processes and multicomponent matrices systems. Full article

This paper addresses the set stabilization problem of probabilistic Boolean control networks (PBCNs) via a self-triggered control strategy. Firstly, the algebraic representation of the considered PBCNs is established by employing the semi-tensor product (STP) of matrices. Secondly, Lyapunov functions (LFs) are introduced for the set stabilization analysis, and a constructive algorithm for deriving such LFs is provided. On this basis, a necessary and sufficient condition in terms of the LF is obtained to determine whether a PBCN can achieve stabilization to a prescribed target set with probability one. Furthermore, a design method for self-triggered controls (STCs) is developed. Finally, the Escherichia coli lactose operon is presented as an example to validate the theoretical results of this paper. Full article

Waste cooking oil (WCO) represents both a growing environmental burden and a promising low-cost feedstock for sustainable biodiesel production. This review critically examines recent advances in the conversion of WCO into biodiesel, with particular emphasis on heterogeneous catalysts derived from waste resources and their integration within a circular bioeconomy. Waste-derived catalysts obtained from eggshells, snail shells, spent coffee grounds, fish and animal bones, chicken waste, and mineral residues such as marble are comparatively assessed in terms of biodiesel yield, reaction severity, recyclability, feedstock tolerance, and sustainability. Several Ca-rich waste-derived catalysts achieve biodiesel yields approaching 90–98%, demonstrating their potential to replace conventional homogeneous catalysts while simultaneously valorizing secondary waste streams. However, the analysis shows that maximum biodiesel yield alone is insufficient for identifying the most sustainable catalyst, because high-temperature calcination, alcohol demand, catalyst deactivation, and limited recyclability can offset apparent performance advantages. Particular attention is therefore given to catalyst recovery and reuse, free-fatty-acid-dependent process selection, and the integration of esterification and transesterification routes for variable WCO feedstocks. The review further evaluates biodiesel performance and emissions and critically examines life-cycle assessment (LCA) and techno-economic assessment (TEA). Available LCA evidence indicates that WCO-derived biodiesel can exhibit substantially lower carbon and cumulative energy burdens than first-generation biodiesel, although outcomes remain strongly dependent on system boundaries, allocation procedures, energy sources, catalyst preparation, and avoided-waste credits. By integrating feedstock pretreatment, waste-derived catalyst selection, biodiesel conversion, engine performance, LCA, and TEA within a unified framework, this review identifies the major research gaps and provides a pathway toward scalable, low-carbon, and economically viable WCO valorization. Full article

Improving pre-harvest sprouting (PHS) resistance in white-grained wheat is challenging because reduced grain pigmentation is often associated with weaker seed dormancy. Here, we evaluated the breeding value of the white-grained, PHS-resistant landrace Tuotuomai (TTM) and developed a molecular marker for tracking TTM-derived material during breeding. Based on phenotypic segregation and sequence polymorphism analysis, a KASP marker targeting a polymorphic site in the TaMyb10-D1 region was developed. The marker co-segregated with white grain color in the TTM × CS population, and the TTM-type marker genotype occurred more frequently in low-germination-percentage (GP) than in high-GP groups of the TTM × AK58 segregating populations. The marker was subsequently combined with phenotypic selection in pedigree and multi-parent breeding schemes. The derived white-grained line SMBK3 had a GP of 29.33%, compared with 98.67% for AK58, and showed reduced plant height and increased thousand-kernel weight relative to TTM. The multi-parent-derived line SMBK4 combined white grain with a GP of 46.93%, compared with 100% for the susceptible white-grained parents. Analysis of SMBK3-derived lines indicated that additional genetic factors, including the PHS1 haplotype, also contributed to variation in GP. These results demonstrate the breeding value of TTM and support the use of the TaMyb10-D1-linked KASP marker as a tracking tool, together with phenotypic selection, in TTM-derived white-grained wheat breeding. Full article

Background: Schizophrenia is frequently underdiagnosed or diagnosed late due to the lack of objective diagnostic markers. This study aimed to develop and validate a machine learning model using routine hematological parameters for the auxiliary diagnosis of schizophrenia. Methods: A total of 150 first-episode drug-naïve schizophrenia patients and 150 healthy controls were enrolled. Study parameters included routine hematological parameters and its derived inflammatory markers. Feature selection was performed using Elastic Net regression, followed by the construction of an L2-regularized logistic regression model. Model discriminative performance, calibration, and clinical utility were assessed through internal validation, area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis. An independent cohort of 50 schizophrenia patients and 50 patients with major depressive disorder (MDD) was used for exploratory differential diagnostic evaluation. Results: The final model incorporated 12 features: neutrophil count (NEUT), eosinophil count (EO), mean platelet volume (MPV), hematocrit (HCT), hemoglobin (HGB), neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), systemic immune-inflammation index (SII), systemic inflammation response index (SIRI), neutrophil-to-platelet ratio (NPR), along with age and sex. The model achieved an AUC of 0.859 (95% CI: 0.780–0.937) on the test set, with an accuracy of 0.767, sensitivity of 0.711, and specificity of 0.822. Calibration curves confirmed good calibration, and decision curve analysis further verified its clinical utility. In the exploratory differential diagnostic analysis, the model showed limited performance in distinguishing schizophrenia from MDD. Conclusions: The model based on routine hematological parameters and L2-regularized logistic regression can effectively differentiate first-episode drug-naïve schizophrenia patients from healthy controls, providing a low-cost and easily accessible auxiliary diagnostic tool for clinical practice. Full article

The growing emphasis on environmental, social, and governance (ESG) performance has increased the need for structured decision-support approaches that can assess sustainability maturity while accounting for both operational efficiency and broader sustainability priorities. However, existing sustainability maturity models are often static, sector-specific, or limited in their ability to integrate expert knowledge with data-driven analytical support. In this study, sustainability maturity refers to the extent to which ESG principles and practices are systematically embedded in organizational processes and decision-making, rather than merely reflecting current ESG performance or an aggregate ESG score. This study proposes an integrated sustainability maturity assessment framework that combines lean–green sustainability principles, ESG criteria, Fuzzy Analytic Hierarchy Process (Fuzzy AHP), and exploratory Random Forest analysis, linking expert-based criterion weighting with maturity assessment and feature-level interpretation within a common decision-support architecture. The framework evaluates sustainability maturity across ten industrial sectors using 33 ESG-oriented sub-criteria structured under environmental, social, and governance dimensions, derived from the sustainability literature and relevant standards and frameworks and refined through expert consultation. Twenty-four experienced professionals contributed to the assessment across the ten sectors, which were selected to reflect diverse sustainability practices and sector-specific conditions. Fuzzy AHP is employed to derive the relative importance of the criteria from expert judgments, while the resulting weighted assessment structure is used to determine sector-level sustainability maturity. Random Forest analysis is subsequently applied to the same weighted ESG criteria, using the resulting maturity classifications as target classes, to explore maturity-related patterns and identify influential criteria based on Gini impurity-based feature importance. The results show clear cross-sector variation: Information Technology (0.87) and Energy (0.84) achieved Level 5 maturity, whereas Food (0.55), Textile (0.48), and Logistics (0.46) were classified at Level 3. Regulatory Compliance Initiatives (0.109) and Certification Continuity (0.098) showed the highest feature importance. By integrating expert-based weighting, ESG maturity assessment, cross-sector benchmarking (i.e., comparison of sectors using the same weighted ESG assessment structure), and exploratory machine-learning-based feature prioritization within a single decision-support architecture, the proposed framework enables organizations to identify maturity gaps and prioritize sustainability improvement areas. Rather than serving as a deterministic predictive model, the framework is intended as an adaptable analytical and managerial decision-support tool for sustainability assessment and strategic planning across diverse industrial contexts. Full article

Jobs–housing balance is an important goal of urban sustainable development, and Transit-Oriented Development (TOD) is widely regarded as an effective pathway to optimize the jobs–housing relationship. However, station-area-scale jobs–housing studies remain limited, and existing jobs–housing typologies have not been linked to regulable TOD characteristics, leaving jobs–housing optimization without a basis for differentiated regulation. Using 72 built subway station areas in Shenzhen’s Density Zone 1 as samples, this study draws on three dimensions, namely transport supply (Node), land use (Place), and jobs–housing, and applies hierarchical clustering, multiple linear regression, and Lasso regression to examine jobs–housing typologies, land use and building distribution characteristics, and the effects of transport supply and land use on the jobs–housing relationship in TOD station areas. Regression analysis uses jobs–housing value, a weighted score derived from the employment–residential area ratio (JH1) and the employment–residential population ratio (JH2), as the dependent variable. The main findings are as follows. (1) TOD station areas can be classified into four types, which differ significantly in land use and building distribution; planning strategies should therefore be differentiated according to the characteristics of each type. (2) The four types exhibit a concentric spatial structure comprising an employment-oriented core, a jobs–housing balanced middle ring, and a residential-oriented periphery, with the mean betweenness centrality increasing gradually, indicating that TOD intensity rises in tandem with subway network hub status. (3) Across the full sample, the interaction term between transport supply and land use is significantly and positively associated with jobs–housing value, a result supported by spatial econometric robustness checks; the main effects of Node and Place are significantly negative only in ordinary least squares. (4) Subway line direction, bus stop density, floor area ratio, and maximum planned floor area ratio are positively associated with jobs–housing value, whereas shared bike density, POI density, total road length, and building mixing entropy are negatively associated with it. This study provides a quantitative basis and actionable planning pathways for the differentiated regulation of the jobs–housing relationship in TOD station areas. Full article

This study evaluated the baseline performance of unmodified Moringa oleifera seeds and crustacean-derived chitosan for uranium removal from aqueous solutions relevant to naturally occurring radioactive material (NORM) contamination. Batch adsorption experiments were conducted using uranium-spiked solutions with nominal concentrations of 10, 50, and 500 µg/L (measured baseline concentrations of 8, 38, and 484 μg/L). A Taguchi L27 fractional factorial design was applied to optimize uranium biosorption parameters for M. oleifera seeds, while chitosan was assessed through baseline comparative batch trials. All batch experiments, including biosorbent-free controls, were conducted in triplicate to evaluate their reproducibility. Chitosan exhibited limited uranium removal, with a maximum of 34.68% at 8 μg/L. It also formed viscous suspensions that hindered filtration. In contrast, M. oleifera seeds demonstrated superior adsorption performance, achieving a maximum removal efficiency of 99.32% and a maximum adsorption capacity of 0.93 mg/g (932 µg/g). Analysis of means (ANOM) and analysis of variance (ANOVA) identified initial pH and initial uranium concentration as the co-dominant factors governing removal efficiency. The highest removal efficiency was obtained at 484 µg/L, pH 4, 25 °C, 2.0 g/L adsorbent dosage, and 90 min contact time. Adsorption isotherm modeling derived from dedicated equilibrium trials showed strong agreement with the Freundlich model (R² > 0.95), consistent with sorption onto energetically heterogeneous surface sites. Overall, unmodified M. oleifera seeds show high baseline efficacy for dilute uranium removal. However, the optimum conditions identified here are more acidic than the near-neutral pH of most NORM-impacted waters; thus, practical implementation must also account for competing carbonate speciation. Full article

Digital twin (DT) technology is increasingly mobilized to improve the efficiency and operational performance of renewable energy (RE) systems. This study maps the efficiency-oriented segment of DT research in RE through a bibliometric analysis of 383 documents indexed in Web of Science and Scopus (2018–September 2026), obtained after screening out records in which the acronym DT denotes another concept, and compares the results with two earlier versions of the corpus. Research gaps are derived through a three-level triangulation of keyword prevalence, co-occurrence cluster composition, and position on the strategic diagram; they are tested across keyword thresholds, clustering algorithms, corpus subsets, and 246 runs of the strategic diagram, and are interpreted as gaps in salience within the analyzed corpus rather than as proof of absence from the wider literature. Annual output roughly doubled each year from 2021 to 2025 under every growth estimator, with China being the leading contributor. Machine learning and energy management form the most developed themes, whereas economic appraisal remains marginal: no economic term reaches the keyword core, no theme is organized around an economic construct in any run, and the two economic magnitudes reported by the most cited documents never set the cost of the twin against its benefit. Three gaps are retained—the absence of standardized appraisal frameworks for the twin itself, the weak consolidation of interoperability research, and the scarcity of work at the integrated, multi-energy scale—together with the peripheral coverage of hydropower and retrofit. Four stakeholder-specific recommendations follow, each linked to its evidence and to an existing practical precedent. Full article

Industrial production systems require intelligent maintenance solutions to minimize unplanned downtime, improve reliability, and support data-driven decision making. This study presents a machine learning framework for condition monitoring and fault classification in industrial production systems. The framework is evaluated using an AI4I-derived synthetic dataset comprising 10,000 production events characterized by operational sensor measurements and machine failure indicators. An exploratory analysis is first conducted to examine data distributions, failure patterns, and relationships among operational variables. Subsequently, four classification models (Logistic Regression, Random Forest, Histogram-Based Gradient Boosting, and a Multilayer Perceptron (MLP) neural network) are developed and comparatively evaluated. The results show that nonlinear models significantly outperform the Logistic Regression baseline, with Random Forest, Histogram-Based Gradient Boosting, and MLP achieving very high classification performance. The findings suggest that interactions among operational variables contribute substantially to the fault classification task and are more effectively captured by nonlinear learning approaches than by linear models using the original feature set. Overall, the study provides a benchmark-style comparative evaluation of representative machine learning classifiers for fault classification on an AI4I-derived synthetic dataset. The findings primarily illustrate classifier behavior under controlled synthetic conditions and provide a basis for future validation using real industrial data and operational maintenance environments. Full article

Background: Peripheral nerve injury causes substantial functional impairment and represents a major health concern worldwide. Peripheral nerve injuries are sequelae of iatrogenic injuries, occupational trauma, and motor vehicle crashes. The available therapeutic approaches, including pharmacological treatment, surgical repair, and rehabilitation, are mostly management therapies rather than curative treatments. Consequently, functional recovery is never achieved. There is a dire need for new therapeutic approaches to facilitate functional restoration. Kaempferol-3-O-rutinoside (K3R) is a plant-derived flavonoid glycoside known for its potent anti-inflammatory and antioxidant properties. Flavonoids may promote tissue recovery following injury by modulating inflammatory responses and reducing oxidative stress. However, the effect of K3R on functional recovery following peripheral nerve injury has not yet been investigated. Therefore, this study aimed to evaluate the effects of K3R on early functional recovery following sciatic nerve crush injury in mice. Methods: Twenty-four BALB/c mice were equally allocated to three groups (n = 8): (1) Sham, (2) Ctrl (Control), and (3) K3R treatment group. K3R (2.5 mg/kg) was dissolved in DMSO and administered intraperitoneally once a day from the time of sciatic nerve injury until the conclusion of the experiment. The Ctrl and sham groups received an equal volume of DMSO according to the same dosing schedule. Sensorimotor function recovery was assessed using behavioral tests, including grip strength, pinprick, sciatic functional index (SFI), and hot plate tests. Hematological and biochemical assays were performed to assess total blood count and oxidative stress markers. Muscle fiber morphology was evaluated by morphometric analysis. Proinflammatory cytokines (TNF-α and IL-6) were measured as indicators of the systemic inflammatory response, whereas S100β protein levels were measured as a marker associated with neural injury and glial responses following sciatic nerve injury. Results: K3R treatment significantly enhanced early sensorimotor function recovery compared with the Ctrl group (p < 0.001). Significant differences in oxidative stress markers TAC (p < 0.001), TOS (p < 0.001), MDA (p = 0.003), and CAT (p = 0.007) were observed among the experimental groups. When comparing the K3R-treated group to the Ctrl group, morphometric analysis revealed an increase in the diameter of muscle fibers (p = 0.011). The levels of IL-6 (p = 0.042) and TNF-α (p = 0.005) indicated attenuation of the systemic inflammatory response, while reduced S100β levels (p = 0.002) were observed in the K3R-treated group, consistent with reduced injury-associated changes. Conclusions: Our findings suggest that K3R may facilitate early functional recovery following sciatic nerve injury, potentially by attenuating oxidative stress and systemic inflammatory responses. However, the precise mechanisms underlying this remain to be elucidated. Future studies should therefore focus on identifying the molecular pathways involved in K3R-mediated recovery, as well as axonal regeneration, Schwann cell activity, and remyelination. Long-term investigations incorporating structural, morphometric, and molecular assessments will be essential to understand its regenerative efficacy and translational potential. Full article

Background/Objective: Co-exposure to heavy metals from environment and a high-fat diet (HFD) represents a growing health concern, contributing to various diseases, but their combined effects on brain health remain less explored. This study aimed to evaluate the neuroprotective potential of Shogaol-enriched ginger extract (SEGE) against heavy metals- and HFD-induced neuropathology. Plant-derived isolated pure compounds can be costly and less accessible; SEGE may offer a translatable, multi-target dietary intervention. Methods: Male Balb/c mice (8–11 weeks old) were exposed to a metal mixture of arsenic (As), lead (Pb), and aluminum (Al; 25 mg/kg/day each) in drinking water and 40% HFD in feed for 60 days. SEGE was administered via feed at two different doses (2 mg/kg/day and 12 mg/kg/day). Assessments including gene expression analyses (using Quantitative Reverse Transcription Real Time Polymerase Chain Reaction (qRT-PCR)) of synaptic and cholinergic markers, spectrophotometric measurement of acetylcholine (ACh) levels, neuronal counting in the cortex and hippocampus via histology, and Gas chromatography mass spectrometric (GC/MS) analysis for brain metabolic quantification were performed. Results: The combined toxic exposure significantly downregulated the expression of synaptic plasticity markers (Synaptophysin, Polysynaptic Density Protein 95 (PSD95), and Calcium/Calmodulin-Dependent Protein Kinase-IV (CAMK-4)) and α and β Nicotinic Acetylcholine Receptors (α7nAChR, α4nAChR, and β2nAChR) in the hippocampus and cortex. ACh levels were also reduced along with significant neuronal loss in the cortical layers and the hippocampal regions. Met + HFD disrupted the brain’s metabolic profile. SEGE treatment significantly restored these markers and attenuated the cholinergic deficits. SEGE treatment, specifically at a higher dose, demonstrated a rescuing effect on the brain’s metabolic profile when compared to the metals and HFD. SEGE further preserved the neuronal count at a higher dose (12 mg/kg/day), exhibiting superior protective effects. Conclusions: These findings suggest that SEGE mitigates neurotoxicity induced by the interaction of toxic metals and dietary factors, likely by preserving synaptic plasticity, metabolic profile, and cholinergic functions, particularly at a higher dose. Full article

Different approaches to skeletal muscle regeneration using progenitor cells are extensively studied as strategies for the treatment of muscle tissue pathologies. Recently, it was discovered that anatomically localized alveolar mucosa multipotent mesenchymal stromal cells (AMCs) are characterized by myogenic potential and high feasibility for muscle tissue recovery and regeneration. Although the resulting multinuclear myotubes express skeletal muscle-specific markers (skeletal myosin, actin, myogenin, and MyoD1), the exact molecular mechanism controlling myogenic differentiation of AMCs is still unclear and poorly scrutinized. In this research, we used a combination of large-scale transcriptome analysis and a bioinformatics approach to investigate molecular pathways and crucial nodes responsible for myogenic differentiation. We studied differentiation of AMC cells in 2D and 3D culture conditions and found core genes with significant expression changes during differentiation and compared them with skeletal muscle-derived stromal cells (SMCs). It appeared that differentiation of AMCs in 3D is significantly different from that in 2D. Moreover, differentiation of AMCs in 2D is closer to differentiation of SMCs in 2D than to AMCs in 3D. Unique properties of AMC differentiation in 3D may be attributed to a significant inhibition of the component of the PDGFRβ signaling pathway responsible for ruffle organization. Full article

Post-traumatic stress disorder (PTSD) is associated with immune and autonomic disturbances, but molecular programs linking PTSD-related blood transcriptional signals with autonomic–cardiac biology remain incompletely characterized. The public Gene Expression Omnibus (GEO) cohorts analyzed here did not directly measure palpitations or autonomic dysfunction. Peripheral-blood transcriptomes from GSE81761 and GSE63878 were integrated with a prespecified pathway-derived autonomic–cardiac gene set constructed from Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Genes overlapping the curated pathway set were assessed using functional enrichment and protein–protein interaction (PPI) analyses. Multi-algorithm comparison and a separate random-forest/SHapley Additive exPlanations (SHAP) analysis were performed for internal model assessment and biological prioritization. GSE64813 and GSE97356 were analyzed using targeted single-gene analyses and cohort-specific multivariable logistic models. Model performance was examined using repeated nested cross-validation, a fully nested sensitivity analysis, and locked-transfer testing to GSE64813 and GSE97356. Interleukin 15 (IL15) and protein phosphatase 2 scaffold subunit Aalpha (PPP2R1A) were evaluated using external-cohort analyses, immune-cell deconvolution, and descriptive postmortem hippocampal single-nucleus data. Among 1679 nominally significant PTSD-associated candidate genes, 92 overlapped the curated pathway set and were enriched for cytokine signaling, chemotaxis, apoptosis, calcium transport, and PP2A-related functions. PPI analysis yielded 37 recurrent candidate hub genes. The original model comparison ranked support-vector machine (SVM) the highest across the merged and source-cohort summaries [mean area under the receiver-operating-characteristic curve (AUC) 0.853], but these estimates represent internal discovery-stage performance. IL15 was recurrently prioritized by network-based analyses, whereas PPP2R1A was a component of an enriched phosphatase-related module. IL15 and PPP2R1A showed modest single-gene effects in GSE64813 (AUC 0.566 and 0.612) and GSE97356 (AUC 0.558 and 0.593). Cohort-specific refitted models showed apparent AUC values of 0.834 and 0.749, respectively. Repeated nested cross-validation conditional on the preselected 29-gene feature set identified L2-regularized logistic regression as the best-performing algorithm (mean AUC = 0.690). Importantly, a more stringent fully nested sensitivity analysis, in which differential-expression screening and pathway intersection were repeated within each outer training fold, yielded a mean repeated outer-cross-validated AUC of 0.593 [standard deviation (SD) = 0.033; range, 0.541–0.625], indicating limited predictive performance. Locked-transfer AUCs were 0.664 in GSE64813 and 0.534 in GSE97356. Single-nucleus summaries suggested donor- and nucleus-type-dependent expression patterns for IL15 and PPP2R1A. These findings identify IL15-related cytokine signaling and PPP2R1A-related phosphatase regulation as candidate molecular programs linking PTSD-associated transcriptional variation with pathway-derived autonomic–cardiac biology. The results should therefore be interpreted as hypothesis-generating rather than direct molecular evidence for unmeasured palpitation symptoms. Prospective validation in independent, clinically well-characterized cohorts with standardized autonomic and cardiac phenotyping is warranted. Full article

Parkinson’s disease (PD) causes motor control deficiencies, resulting in gait irregularities such as shorter strides, slower walking speed, and irregular step timing. This study presents a video-based deep learning approach for PD classification that extracts skeletal keypoints from Timed Up and Go (TUG) test videos using AlphaPose and the COCO-17 representation. A total of 24 features were generated, comprising 23 conventional gait features and one Freezing of Gait (FoG) feature derived from frequency-domain analysis of ankle velocity signals. This FoG feature was not validated against clinician-confirmed FoG episodes and should be interpreted as a frequency-domain proxy rather than a diagnostic measure. Three feature selection procedures and four LSTM-based architectures were evaluated across full, walking, and turning segments. Experimental results on a self-collected TUG dataset showed that the Standalone FI configuration achieved the numerically highest test accuracy of 77.78% on the turning segment among the evaluated LSTM configurations, while conventional features achieved 66.67% on both the full and walking segments. These test-set metrics provide descriptive estimates derived from a static subject-level test division involving six held-out participants (excluded from training and validation) and should not be viewed as statistically dependable indicators of clinical performance at the population level; in addition, gait-cycle boundaries were not independently validated and fallback usage was not quantified. Turning segments demonstrated higher discriminative power than straight-walking segments. Zero-shot cross-dataset evaluation on Turn-REMAP and PD-Walk revealed a substantial generalization gap, with accuracy falling to 55.12% and 49.53%, respectively, indicating that the present model is not yet suitable for cross-site clinical deployment without adaptation or calibration. The proposed framework provides systematic insights into the comparative role of conventional and FoG-derived gait parameters for non-invasive video-based PD screening. Full article