Search Results (458)

Multiple faults in industrial processes may occur simultaneously or successively, leading to abnormal responses distributed across coupled process regions and evolving over time. This paper proposes a Mamba-based multiple-fault detection framework, termed MFD-Mamba, for dynamic industrial processes. A hybrid multiblock decomposition first combines lag-aware data relationships with prior process connections to construct soft variable memberships and a block-relation prior. The resulting blockwise sequences are then modeled by a shared modified Mamba encoder, while dynamic block relation modeling is used to describe interactions among process regions. Blockwise process predictions are further used to construct hybrid residuals. For fault detection, block residuals are converted into probabilistic evidence and fused within a Bayesian framework, where multiple residual-based monitoring statistics, persistence information, and cross-evidence terms are jointly incorporated to construct the final monitoring index. The proposed method is evaluated on the Tennessee Eastman process and the activated sludge process under overlapping and sequential multiple-fault conditions. On the TE process, MFD-Mamba achieves FDR/FAR values of 99.88%/0.13% and 99.81%/0.20% for the overlapping and sequential fault scenarios, respectively. On the ASP, the corresponding FDR/FAR values are 99.79%/0.78% and 99.58%/1.56%. These results demonstrate that MFD-Mamba provides accurate and reliable multiple-fault detection across different dynamic industrial processes. Full article

Background/Objectives: Antimicrobial resistance threatens public health globally, and critical access hospitals (CAHs) serving rural communities face structural barriers to implementing antibiotic stewardship programs (ASPs). This study evaluated whether a regional educational intervention could improve antimicrobial stewardship knowledge and implementation among rural healthcare professionals. Methods: A quantitative, quasi-experimental repeated cross-sectional evaluation was conducted following Antibiotic Stewardship conferences in 2023 and 2024. Participants from 38 rural and critical access hospitals completed pre-conference, post-conference, 6-month follow-up, and implementation surveys. Quantitative data were analyzed using descriptive statistics and unpaired chi-square tests, with Fisher’s exact test substituted where expected cell counts were below 5, with significance set at p < 0.05. Results: Seventy healthcare professionals participated across both years. Correct identification of the use of an antibiogram to guide empiric (rather than definitive) therapy improved from 27% pre-conference to 58% post-conference (p = 0.029). Confidence in antibiogram interpretation increased from 67% to 100% post-intervention (p = 0.004, Fisher’s exact) but declined to 50% (5 of 10) at 6-month follow-up. Knowledge of the minimum isolate threshold required to construct an antibiogram improved from 23% (7 of 30) pre-conference to 64% (7 of 11) at 6-month follow-up (p = 0.026, Fisher’s exact). Despite this, prescribing intent for conditions where antibiotics are typically unnecessary showed minimal improvement: recommendations for antibiotic use in middle ear infections decreased from 50% to 36.84%, and for mpox from 20% to 15.79%. All participants reported intent to modify clinical practice; however, time constraints, technology limitations, and lack of resources were consistently cited as primary barriers for implementation across both survey years. Conclusions: A regional educational intervention improved short-term antimicrobial stewardship knowledge and confidence among rural healthcare professionals, but knowledge retention declined over time and structural barriers persisted. Education alone is insufficient; sustained reinforcement and rural-adapted ASP models incorporating protected ASP time, decision-support tools and external partnerships are needed to translate knowledge gains into consistent stewardship practice. Full article

Background/Objectives: In the search for potent non-sugar α-glucosidase inhibitors with improved safety profiles, a novel flavonoid glycoside was isolated for the first time from the stem bark of Albizia saponaria (Fabaceae). The objective of this study was to elucidate its chemical structure and evaluate its therapeutic potential as an anti-hyperglycemic agent compared to known related flavonoids and a standard clinical drug. Methods: Comprehensive structural elucidation was performed using high-resolution mass spectrometry and multidimensional 1D/2D NMR (¹H, ¹³C, HSQC-DEPT, COSY, and CIGAR). To assess its inhibitory efficacy and pharmacokinetic profiles, an in silico comparative study was conducted against a database of related flavonoids (Quercitrin, Hyperoside, and Isoquercitrin) and the clinical drug Acarbose. This involved molecular docking simulations against human intestinal maltase-glucoamylase (PDB ID: 3TOP) alongside integrated ADMET modeling and toxicological screening. Results: The compound was successfully identified as 4′,7-dihydroxyflavan-3′-O-β-D-glucoside (1). Molecular docking revealed that Compound 1 exhibited a superior predicted binding affinity of −9.5 kcal/mol, outperforming Quercitrin (−9.3 kcal/mol), Hyperoside (−8.3 kcal/mol), Isoquercitrin (−7.9 kcal/mol), and Acarbose (−7.2 kcal/mol). This strong thermodynamic stability is driven by a robust conventional hydrogen-bonding network with key active site residues (Arg1377, Gln1372, and Gly1365), successfully overriding a localized electrostatic strain at Asp1279. Furthermore, ADMET modeling demonstrated a highly desirable local pharmacokinetic framework; its low Caco-2 permeability (−6.432) and low human intestinal absorption (HIA = 0.120) favor targeted luminal retention in the gastrointestinal tract, mirroring Acarbose while minimizing systemic exposure. Crucially, toxicological screening unveiled a significant safety advantage for Compound 1, marked by negligible CYP3A4 interaction (0.004) and a remarkably low risk of Drug-Induced Liver Injury (DILI = 0.213) compared to the high-risk hepatotoxic profile of Acarbose (DILI = 0.882) and the reference flavonoids (DILI > 0.69). Conclusions: These predictive findings establish Compound 1 as a highly promising, low-toxicity natural scaffold for anti-hyperglycemic drug development. Its superior binding affinity and minimized hepatotoxicity risk warrant subsequent in vitro and in vivo functional validation. Full article

Wheat (Triticum aestivum), a key global crop, faces rising drought stress that limits root growth and water uptake. Root meristem growth factors (RGFs) are small peptides that regulate root stem cell maintenance, meristem activity, and lateral root formation in model plants, yet the RGF gene family remains unexplored in wheat. Here, we performed a comprehensive genome-wide analysis of the TaRGF gene family, identifying 15 genes distributed across the A, B, and D subgenomes and classified into five homeologous groups (TaRGF1–TaRGF5), predominantly located on chromosomes 2 and 6. All TaRGFs contained a characteristic RGF motif, with dibasic cleavage sites and Asp–Tyr motifs indicating conserved maturation mechanisms. Based on the phylogenetic analysis, the TaRGF5 homeologs showed the highest similarity to Arabidopsis thaliana RGF5. Tested RNA-seq data revealed predominantly root-enriched expression for all TaRGF genes, with TaRGF5 exhibiting the most root-preferential and downregulation under drought stress. Quantitative real-time PCR (qRT-PCR) confirmed that drought stress suppressed the expression of TaRGF5A, TaRGF5B, and TaRGF5D in roots of wheat cultivar Sids-13 across all time points, unlike the higher accumulation seen in controls. Promoter analysis predicted a unique BES1 transcription factor binding site exclusively in TaRGF5B, linking brassinosteroid signaling to peptide-mediated root regulation. Structural modeling and molecular docking predicted an interaction between wheat TaRGF5 homeologs and root growth factor-insensitive receptor kinase (TaRGI3), characterized by conserved sulfotyrosine-mediated binding and favorable interaction energetics. Based on this characterization of the wheat RGF gene family, particularly the potential role of TaRGF5 in root development and drought-adaptation signaling, we propose targeting this gene for functional analysis to improve wheat resilience under water-limited conditions. Full article

Colorectal cancer is a leading cause of cancer death, yet its molecular heterogeneity remains poorly translated into individualized treatment. We present a reproducible artificial intelligence (AI) framework that integrates multi-omics benchmarking, sample-level drug prioritization, E3 ubiquitin ligase selection, and shape-anchored Proteolysis Targeting Chimera (PROTAC) design for KRASG12D in colon adenocarcinoma (COAD). A controlled synthetic benchmark comprising 425 tumor and 41 simulated normal profiles, parameterized to match The Cancer Genome Atlas (TCGA) distributions, was used for pipeline verification. Among sixteen methods, the Balanced Latent Integration with Stability Selection (BLISS) model achieved the highest silhouette width (0.86) and competitive agreement (Adjusted Rand Index, ARI, 0.90). The pipeline was validated on real data: a TCGA COAD cohort (186 tumors) with independent Consensus Molecular Subtype (CMS) labels and a CPTAC cohort (104 tumors). Integration modestly recovered CMS (ARI 0.28), and stage, not molecular cluster, drove survival (log-rank p = 0.005 versus 0.81). Sample-level prioritization differed from cluster-level ranking in 82.6% of profiles, below chance (p < 0.0001), without indicating efficacy. Candidate NOVEL00489 showed a good MM-GBSA estimate, matching the reference ASP3082. Compounds are computational candidates requiring experimental validation. This establishes a transparent benchmark for in silico degrader generation in precision oncology. Full article

The long-term accumulation of polyethylene terephthalate (PET) in marine environments may drive the evolution of microbial degradation capabilities, positioning the ocean as a valuable reservoir for discovering novel PET-degrading microorganisms. In this study, we isolated 305 marine-derived microorganisms with potential PET-degrading capability from samples collected from mangrove areas of Zhanjiang and the intertidal zones of Daya Bay, Shenzhen, China, using PET powder as a major carbon source. Subsequent evaluation of degradation performance via scanning electron microscopy and Fourier-transform infrared spectroscopy analysis identified 14 isolates capable of degrading PET film. These 14 strains belonged to 14 distinct species, none of which, to the best of our knowledge, has been previously documented as PET degraders. Among them, Microbacterium aurum SCSIO 85700 exhibited the most potent PET-degrading activity, achieving a weight loss of 2.1 mg (2.1%) and a 6.5% increase in relative crystallinity over 30 days. Genome analysis revealed the genetic basis underlying PET degradation and associated metabolic pathways in strain SCSIO 85700. Notably, genome mining and structural modeling identified two candidate polyester hydrolases, MA2267 and MA2443, possessing conserved His–Asp–Ser catalytic triads and exposed substrate-binding clefts resembling those of characterized PET-degrading enzymes, suggesting their potential involvement in PET depolymerization. Collectively, this study expands the recognized diversity of marine PET-degrading microorganisms and provides microbial resources for sustainable PET bioremediation. Full article

This study investigated the RNA virome associated with the phlebotomine sand fly Pintomyia fischeri using a metatranscriptomic approach applied to the Meta29 library, composed of 21 specimens collected at the São Paulo Zoo, Brazil. Read-based taxonomic analysis revealed a diverse viral community composed of viruses associated with the families Iflaviridae, Narnaviridae, Partitiviridae, Reoviridae, Solemoviridae, Tombusviridae, Totiviridae, and Tymoviridae, in addition to highly abundant unclassified RNA viruses related to the ShiM 2016 group. Assembly and annotation analyses enabled the characterization of four viral RNA genomes associated with the families Partitiviridae (PfPartitiV-1a-SP, PfPartitiV-1b-SP, and PfPartitiV-2-SP) and Narnaviridae (PfNarnaV-1-SP). Sequence comparisons revealed low amino acid identity relative to currently available reference viral sequences, supporting the classification of these sequences as highly divergent and potentially novel viral lineages. Structural modeling of the RNA-dependent RNA polymerase (RdRp) proteins identified the viral polymerase-specific catalytic motifs A, B, and C. Phylogenetic analyses further supported the evolutionary divergence of the identified viruses relative to currently described taxa. Overall, these findings expand current knowledge of viral RNA diversity associated with Neotropical sand flies and highlight the value of metatranscriptomic approaches for the detection and characterization of previously undescribed viruses associated with medically important insect vectors. Full article

A multi-scale framework is proposed for predicting the equivalent engineering elastic constants of carbon/glass hybrid composite laminates with different fiber volume fractions V_f by integrating multiscale finite element homogenization with analytical micromechanics. Biaxial carbon fiber fabric reinforced vinyl ester composites are used as the target material system, and representative volume element (RVE) models are constructed sequentially at the lamina, laminate, and hybrid-laminate scales. At the lamina scale, the analytical self-consistent field micromechanics (SCFM) approach and the RVE method give highly consistent predictions for the fiber-dominated longitudinal modulus E₁ and in-plane shear modulus G₁₂, with deviations below 2%. However, the SCFM approach overestimates the transverse modulus E₂ and transverse Poisson’s ratio ν₂₃ as the V_f rises. The unidirectional lamina constants predicted by SCFM and RVE are then used as inputs for an analytical solution for predicting elastic constants (ASPE) and a finite-element-analysis-based micromechanics (FEAM) model of laminate and hybrid-laminate. At the laminate scale, the moduli predicted by ASPE(SCFM) and ASPE(RVE) differ by less than 2%, whereas the out-of-plane properties are sensitive to the transverse lamina inputs. Experimental validation shows that the E_x values predicted by ASPE(RVE) and FEAM(RVE) deviate from the measured average by approximately 3.5%, with good agreement also obtained for G_xy and ν_xy. At the hybrid-laminate scale, ASPE(RVE) and FEAM(RVE) exhibit only minor differences in estimating the in-plane moduli, out-of-plane moduli, and Poisson’s ratios. Full article

Pseudomonas aeruginosa is a predominant spoilage organism in meat products. The extensive use of antimicrobial preservatives has, however, led to the emergence of resistant strains. Mitigating the resistance of foodborne P. aeruginosa to conventional preservatives has become a critical challenge in food safety. Piperonylpiperazine (Pip), a piperazine derivative sourced from Piper nigrum, was first examined in this work as a novel agent capable of both suppressing virulence and enhancing preservative efficacy against P. aeruginosa, with its mode of action elucidated. At sub-inhibitory concentrations, Pip strongly suppressed the production of virulence factors and potentiated the susceptibility of P. aeruginosa to the common preservative EDTA. Mechanistically, a multi-pronged approach involving pull-down assay, transcriptomic profiling, isothermal titration calorimetry (ITC) analysis, and gene knockout models demonstrated that Pip binds specifically to the LYS-164 and ASP-35 residues of the RhlI synthase, blocking the quorum sensing (QS) signaling cascade. This QS disruption led to reduced virulence factor production and attenuated pathogenicity in a Caenorhabditis elegans model. The compromised QS system subsequently induced oxidative stress, which disrupted cell membrane integrity and permeability, thereby potentiating EDTA’s antibacterial action. These findings suggest that Pip is a promising natural additive that can be used in combination with existing preservatives to enhance food safety. Full article

In silico structural characterization of carbohydrate-active enzymes (CAZymes) in Bacillus subtilis T7 reveals mechanistic insight into the strain’s capacity for consolidated bioprocessing of untreated lignocellulosic biomass. Homology models for 13 CAZymes were constructed using SWISS-MODEL, with Cu²⁺ and FAD cofactors incorporated into the AA10 lytic polysaccharide monooxygenase and AA3 oxidoreductase models, respectively. Blind molecular docking across the full protein surface identified energetically favorable binding pockets on GH9 endoglucanase and Abhydrolase_1. Among 12 enzyme–ligand pairs screened, Abhydrolase_1 exhibited the highest affinity for xylotetraose (−7.7 kcal/mol) and GH9 showed the strongest preference for cellotetraose (−7.0 kcal/mol). Site-specific docking confirmed six hydrogen bonds with Gly32, Phe33, Thr34, Ser36, Arg179, and His256, supplemented by two carbon–hydrogen bonds with Ile180 and Ser39, anchoring xylotetraose within the Abhydrolase_1 binding cavity, and seven hydrogen bonds stabilizing cellotetraose in the GH9 catalytic groove, with key contacts at Tyr141, Trp145, Asp194, Trp193, Arg254, Tyr255, and Tyr354. One-hundred nanosecond all-atom molecular dynamics simulations (GROMACS 2023.2, CHARMM36 force field, triplicate runs) confirmed overall structural integrity for both proteins: Abhydrolase_1 maintained a compact conformation (Rg = 18.11 ± 0.09 Å; backbone RMSD 2–3 Å), while GH9 was similarly stable (Rg = 30.36 ± 0.33 Å; RMSD 2–5 Å). Ligand dynamics were more variable—xylotetraose remained bound within the Abhydrolase_1 active site for approximately 75 ns before partial displacement, whereas cellotetraose exhibited dynamic association along the GH9 catalytic channel, consistent with processive substrate translocation in endoglucanases. These computational findings line up with the strain’s experimentally observed hydrolytic clearance zones (cellulase 24.5 mm; xylanase 11.6 mm), 63.4% alkali lignin decolorization, transient accumulation of ferulic acid and vanillin, and a hydrogen yield of 1.41 mol H₂/mol substrate from untreated food waste. Together they give a molecular-level picture of substrate-specific CAZyme recognition in B. subtilis T7 and support its potential as a pretreatment-free platform for lignocellulosic biohydrogen production. Full article

N-acetylornithine aminotransferase (AcOAT) is a pyridoxal 5′-phosphate (PLP)-dependent enzyme that catalyzes a key transamination step in arginine biosynthesis. In cyanobacteria, arginine metabolism is closely associated with nitrogen assimilation and storage, yet biochemical information on cyanobacterial AcOATs remains limited. In this study, the AcOAT encoded by the cce_3094 gene from Crocosphaera subtropica ATCC 51142 (CsAcOAT) was cloned, heterologously expressed, purified, and systematically characterized. Recombinant CsAcOAT was obtained as a soluble protein with an apparent molecular mass of approximately 43 kDa. Steady-state kinetic analysis showed that CsAcOAT catalyzed transamination between N-acetylornithine (AcOrn) and α-ketoglutarate (α-KG), with apparent K_M values of 0.17 ± 0.03 mM for AcOrn and 0.020 ± 0.003 mM for α-KG, indicating a higher affinity for α-KG. The enzyme exhibited optimal activity at pH 8.5 and 30 °C, retained relatively high activity over a broad temperature range of 0–50 °C, and was activated by Zn²⁺ and Co²⁺ but inhibited by Ni²⁺. Structural analysis based on homology modeling, molecular docking, and molecular dynamics simulations suggested a conserved PLP-dependent aminotransferase fold and a stable binding mode for the PLP-AcOrn complex in the active-site pocket. Site-directed mutagenesis further demonstrated that Gly114, Asp239, Lys268, and Thr296 are indispensable for catalytic activity, whereas Ser113, Ala115, and Gln242 make important contributions to catalytic turnover and cofactor-assisted catalysis. These results provide biochemical and structural characterization of CsAcOAT, expand current knowledge of cyanobacterial AcOATs, and offer a useful basis for future studies on arginine metabolism and nitrogen storage in diazotrophic cyanobacteria. Full article

Deep learning-driven medical image analysis has become a critical auxiliary tool for clinical oral disease diagnosis. However, most existing caries detection methods based on YOLO series models are validated on single-lesion datasets, leading to generally high missed-detection rates and unsatisfactory localization accuracy in complex dental radiographic scenarios where multiple dental conditions and restorations coexist. To address this limitation, this paper proposes YOLO-ASPE, an improved YOLOv11n lightweight framework for tiny-carious-lesion detection. It integrates a P2 high-resolution detection branch, a multi-scale sequential feature fusion module, a lightweight SE channel attention mechanism, and EIoU bounding box regression loss. Experiments were conducted on a public eight-category dental radiographic dataset with caries as the sole detection target. The results show that YOLO-ASPE achieved a Caries AP@0.5 of 68.7%, 10.5 percentage points higher than the baseline model, with a real-time inference speed of 312 FPS. This work provides a lightweight technical reference for the computer-assisted screening of dental caries, although further external validation is still required before clinical application. Full article

Objective: To investigate the clinical characteristics, inheritance patterns, and genotype–phenotype correlations of DHX37 variants in 46,XY disorders of sex development (DSD). Methods: We retrospectively reviewed 108 patients with 46,XY karyotype who underwent DSD evaluation and trio-based whole-exome sequencing (trio-WES) at our center between January 2021 and December 2025. Six probands with DHX37 variants and no concurrent pathogenic or likely pathogenic variants in other known DSD-associated genes were analyzed in detail. Clinical presentations, endocrine profiles, imaging findings, and pedigree data were collected. Variant segregation was confirmed by Sanger sequencing; variants were assessed using in silico prediction, conservation analysis, and structural modeling, and were classified according to ACMG/AMP guidelines. Results: The six probands exhibited marked phenotypic heterogeneity, with manifestations ranging from complete gonadal dysgenesis to mild testicular underdevelopment with gynecomastia. Six heterozygous DHX37 missense variants were identified across distinct functional domains: one in RecA1 [c.1000C>T (p.Arg334Trp)], four in RecA2 [c.1379T>C (p.Val460Ala), c.1432G>A (p.Gly478Arg), c.1730A>G (p.Asp577Gly), and c.1954G>A (p.Val652Ile)], and one in the linker region proximal to the HA2 domain [c.2180C>T (p.Thr727Met)]. Of these, p.Arg334Trp is an established pathogenic variant; p.Gly478Arg has been reported previously, albeit in the same patient included in the present study; and the remaining four were novel. According to ACMG/AMP guidelines, p.Arg334Trp was classified as pathogenic, and the remaining five as variants of uncertain significance. Conclusions: This study provides additional case evidence supporting the pathogenicity of p.Arg334Trp and expands the DHX37 variant spectrum. Computational and structural analyses suggest that p.Gly478Arg may underlie the testicular regression syndrome phenotype of the corresponding proband and that the four novel variants may be involved in testicular development. However, these genotype–phenotype correlations remain speculative; larger cohorts and in vitro functional assays are warranted to confirm these associations. Full article

Objectives: Transient Receptor Potential Ankyrin 1 (TRPA1) has emerged as a stress-responsive ion channel involved in calcium homeostasis, redox signaling, migration, and cancer cell survival; however, the therapeutic relevance of TRPA1 activation versus inhibition remains poorly understood. In this study, we comparatively investigated the anticancer potential of the TRPA1 agonist ASP7663 and the TRPA1 antagonist HC030031 across a broad panel of human cancer cell lines. Methods: Antiproliferative and cytotoxic effects were evaluated using MTT and LDH assays, while apoptotic signaling, DNA fragmentation, mitochondrial membrane potential, migration inhibition, DNA/BSA interactions, topoisomerase I inhibition, and molecular docking analyses were comprehensively assessed. Results: ASP7663 exhibited markedly lower GI₅₀ values than HC030031 in most cancer models and demonstrated favorable tumor selectivity relative to normal cells. Mechanistically, ASP7663 induced robust apoptotic activation characterized by significant upregulation of Caspase-3, Caspase-8, and Caspase-9, together with enhanced DNA fragmentation and pronounced nuclear condensation. Rhodamine-123 staining further revealed substantial mitochondrial membrane depolarization, indicating activation of intrinsic apoptotic pathways. In addition, ASP7663 produced selective membrane damage in malignant cells, stronger inhibition of migration in osteosarcoma and chondrosarcoma models, enhanced CT-DNA/BSA binding affinity, partial topoisomerase I inhibition, and superior docking interactions with apoptosis-related targets, including Caspase-3, Caspase-8, Caspase-9, Bax, and Bcl-2. Conclusions: Collectively, these findings suggest that ASP7663 is a promising multi-target candidate for anticancer therapy and support further investigation of TRPA1-associated pathways as potential therapeutic targets in cancer. Full article

Fused deposition modeling (FDM) is a material-extrusion additive manufacturing technology that is widely used in rapid prototyping, complex product modeling, and functional part fabrication. However, process-parameter drift and environmental disturbances may induce underfilling, overfilling, warping, delamination, and other defects, thereby reducing part quality or interrupting the manufacturing process. Since FDM is characterized by point-wise extrusion and layer-by-layer deposition, layer-surface images naturally contain both spatial morphology and temporal evolution information. Existing image-based diagnostic methods often treat layer images as independent samples, and the selection of the image-stream length is still insufficiently supported by experimental evidence. Moreover, spatial compression in spatiotemporal neural networks may remove local defect information that is important for distinguishing similar process-parameter drifts. This study provides a deployment-oriented analysis of FDM image-stream diagnosis by systematically examining how layer-window length, spatial feature preservation, and strict data partitioning influence process-parameter drift recognition. To address these issues, this paper studies ConvLSTM-based FDM image-stream process-parameter drift diagnosis. Continuous region-of-interest image streams are constructed for one nominal condition and six process-parameter drift conditions. In this paper, the time step T denotes the number of consecutive layer-surface images, or, equivalently, the number of consecutive printed layers, contained in one diagnostic image stream. A ConvLSTM-Flatten baseline is first developed to preserve complete spatial feature maps and to evaluate the effect of different time-step lengths. Then, a ConvLSTM model with adaptive spatial pooling and temporal attention (ASP-TA) is constructed to analyze the influence of spatial pooling granularity and temporal feature fusion. The experiments show that the ConvLSTM-Flatten model achieves the highest average test accuracy of 0.7288 at $T=9$, whereas $T=3$ is identified as a practical optimal time step when test accuracy, image-frame computation, diagnosis latency, and convergence behavior are considered together. The paired trial-wise accuracy difference between $T=9$ and $T=3$ is small and not statistically significant over ten repeated trials. Thus, the diagnostic window corresponding to $T=3$ covers three consecutive deposited layers; after the initial window is available, stride-one stream construction allows the diagnosis to be updated with each newly acquired layer image. ASP-TA with a pooling size of eight consistently outperforms ASP-TA with a pooling size of four, but both are lower than the Flatten baseline, indicating that preserving sufficient spatial information is essential for distinguishing FDM process-parameter drift states. The results reveal the non-monotonic influence of time-step length and clarify the tradeoff between spatial feature preservation and model compactness in FDM image-stream process-parameter drift diagnosis. Full article