Search Results (385)

Background/Objectives: Sleep and circadian disturbances are common in neurodevelopmental conditions, yet the developmental cortical programs linking clock-related transcriptional regulators to disease vulnerability remain unclear. Methods: Here, we integrated human developmental brain transcriptomes, weighted gene co-expression network analysis (WGCNA), human and mouse cortical single-cell atlases, prenatal immune activation transcriptomes, and ASD postmortem brain datasets to characterize the developmental architecture of BrainSpan-derived cortical programs and examine their behavior in perturbational and disease contexts. Results: In the BrainSpan frontal cortex, canonical clock-related genes followed structured but heterogeneous developmental trajectories rather than behaving as a coordinated oscillator-like unit. WGCNA identified a postnatal-rising BrainSpan-derived primary developmental module that was strongly associated with developmental age and enriched for synaptic signaling, neurotransmitter transport, ion transport, membrane excitability, cellular respiration, metabolic regulation, and proteostatic processes. Network analysis placed multiple canonical clock-related and clock-regulatory genes, including NPAS2, BHLHE40, BHLHE41, PER family members, RORA, NR1D1/2, and CLOCK, within a broader neuronal and homeostatic co-expression architecture, although their module-membership strengths varied substantially. Projection onto a human cortical developmental single-cell atlas revealed a non-uniform distribution of the corrected BrainSpan-derived developmental signature, with relatively higher scores in excitatory and inhibitory neuronal populations and lower scores in neuroblast and radial glial populations. A mouse cortical developmental single-cell atlas provided a comparative view of the stage- and cell-type-dependent expression of clock-related genes and the transferred developmental signature during corticogenesis. In a Poly(I:C)-based maternal immune activation dataset, litter-aware reanalysis identified stage-resolved genome-wide transcriptional responses following E12.5 exposure. However, neither the aggregate core clock-gene expression score nor the independently transferred BrainSpan-derived developmental signature showed a significant overall treatment effect or collection-stage-by-treatment interaction, indicating that this bulk dataset provides a perturbational context rather than evidence for selective disruption of the developmental program. An exploratory region-stratified analysis of GSE28521 yielded near-null effects for the BrainSpan-derived developmental signature, with confidence intervals crossing zero across all examined regions. These ASD postmortem findings were therefore treated as a boundary assessment rather than evidence of ASD-specific convergence. Conclusions: Collectively, these findings position clock-related genes as components of a developmentally regulated cortical maturation program enriched for neuronal signaling, synaptic maturation, metabolic regulation, and stress-response processes. Full article

Climate change is intensifying drought, salinity, heat, chilling, flooding, and heavy-metal stresses across major fruit-producing regions, threatening yield stability and fruit quality in economically vital, perennial crops such as apple, grapevine, citrus, banana, strawberry, and peach. Because these species are long-lived, highly heterozygous, and polyploid, conventional breeding for climate resilience remains slow and often inadequate, necessitating molecular strategies informed by systems-level understanding. This review synthesizes recent advances in multi-omics research spanning genomics, transcriptomics, proteomics, metabolomics, epigenomics, ionomics, and phenomics that have collectively decoded the regulatory architecture underlying abiotic stress perception, signaling, and tolerance in fruit crops. Hormonal networks, particularly abscisic acid (ABA) crosstalk with jasmonate, salicylic acid, ethylene, and brassinosteroids, emerge as central integrators of stress responses, coordinating stomatal regulation, osmolyte accumulation, antioxidant defense, and secondary metabolite biosynthesis. Genomic and pangenomic approaches have identified stress-associated loci and cultivar-specific structural variants, while transcriptomic and proteomic studies reveal transcription factor networks (MdERF38–MdMYB1, MaMYB4–MaHDA2, VvDREB1, CsNAC29) and post-translational regulatory switches governing tolerance mechanisms across drought, cold, salinity, and flooding stress. Metabolomic and ionomic profiling link biochemical reprogramming to fruit quality traits, whereas epigenomic mechanisms including DNA methylation, histone modifications, and small RNA regulation provide a chromatin-level layer mediating stress memory across growing seasons. Integration of these omics layers through systems biology, machine learning, and high-throughput phenomics is enabling functional validation via CRISPR-Cas9 and marker-assisted selection, translating correlative associations into causally validated breeding targets. Despite this progress, challenges including batch effects, tissue heterogeneity, and methodological inconsistencies in data integration continue to constrain translational applications. This highlights convergent regulatory hubs across stress types and species, underscoring multi-omics-guided precision breeding as the most promising pathway toward developing climate-resilient, high-quality fruit crop cultivars for sustainable global production. Full article

Cadmium is a persistent toxic metal that promotes oxidative stress, inflammation, and epigenetic dysregulation; however, the research landscape and recurrent microRNA (miRNA)-mediated responses associated with cadmium exposure remain incompletely characterized. This study integrated bibliometric, structured content, and bioinformatics analyses to identify major research trends and recurrent miRNA signals. PubMed/MEDLINE, Scopus, and Web of Science Core Collection were searched using predefined cadmium- and miRNA-related search terms, yielding 1511 source records. After language/document-type filtering and cross-database deduplication, 817 unique records remained for screening, of which 286 publications formed the final bibliometric corpus. Author-keyword co-occurrence analysis was performed using a minimum occurrence threshold of ≥3, with ≥5 used as a sensitivity analysis. Cadmium and microRNA remained the dominant thematic core, while heavy-metal toxicity, apoptosis, and cadmium-stress responses represented the most recurrent secondary themes. In the final human-relevant biological evidence corpus (n = 32 studies), recurrent miRNAs showed substantial directional heterogeneity rather than a uniformly replicated response: miR-155 was reported in four studies with an equal number of increases and decreases, whereas the let-7 and miR-30 families showed modest downward tendencies, and miR-146b and miR-222 showed modest upward tendencies. These small study counts preclude strong claims of directional reproducibility. The miRNA–target network component should therefore be interpreted as exploratory and hypothesis-generating rather than as evidence of cadmium-specific pathway convergence. Overall, this synthesis supports a broader and more reproducible literature map while emphasizing the need for standardized exposure assessment, miRNA measurement, and independent validation of candidate regulatory signals. Full article

In distributed optimization, communication bottlenecks and computational heterogeneity are two critical challenges that undermine system scalability and efficiency. To address these issues, we propose a synchronous algorithm, CC-GT (Conjugate Compressed Gradient Tracking), which integrates conjugate gradient acceleration with dual error-compensated compression to reduce communication overhead and accelerate convergence. Theoretically, we establish its linear convergence under strongly convex and smooth assumptions, and extensive experiments on regression, classification, and neural-network tasks validate the fast convergence and communication efficiency of the synchronous CC-GT. Building on this foundation, we develop GFLCC, an asynchronous extension that eliminates global synchronization barriers while preserving the core mechanisms of CC-GT, thereby inherently mitigating the straggler effect and enabling dynamic load balancing. For the asynchronous GFLCC, extensive comparisons against state-of-the-art asynchronous baselines under harsh network interference demonstrate its strong empirical convergence speed, communication efficiency, and robustness. These results confirm that the proposed framework offers a practical and theoretically grounded solution for communication-constrained and heterogeneous distributed optimization systems. Full article

Distributed network security requires multiple defenders to make coupled resource allocation and coordination decisions under limited budgets, connectivity-dependent risks, and non-negligible coordination costs. We develop a mathematical framework for jointly optimizing defensive resource allocation and coalition formation in a multi-defender network security game. Nodes have heterogeneous values and protection thresholds, attacks propagate through vulnerable connected components, and defenders retain individual budgets while forming coalitions for joint decision-making. For each defender partition, the induced allocation game is evaluated by its worst pure-Nash-equilibrium loss together with a supermodular coordination cost. We prove that optimal protection is NP-hard even in a restricted setting, that every fixed partition induces an exact potential game with finite best-response convergence, and that decentralized equilibria can have an unbounded price of anarchy. We further characterize merge and split thresholds, the piecewise-constant dependence of optimal partitions on the coordination weight, the monotonicity of the selected coordination cost, and conditions for diminishing net merge benefits. Based on these properties, we propose Cost-Aware Coalition Search (CACS), which combines a tractable allocation oracle with multi-start local search over merge, split, and move operations. Experiments on synthetic and real networks show that selective coalitions provide a scalable tradeoff between residual attack loss and coordination cost across different resource levels, coordination weights, and defender populations. The framework provides a mathematical basis for organizing distributed defenders in networked security environments. Full article

Background/Objectives: Lymph-node metastasis remains central to staging, prognosis, surveillance, and treatment planning in malignant melanoma. At present, most statistical models assessing nodal metastatic risk in malignant melanoma consider pathological descriptors, inflammatory markers, metabolic alterations, clinical data, and related variables independently of each other. Therefore, we developed a transparent artificial intelligence (AI)-based approach to assess whether the propensity for nodal metastasis is determined by a single layer of local histopathological aggressiveness or by the integration of different biological levels, including local histopathological aggressiveness, systemic inflammatory–metabolic dysregulation, biological heterogeneity, or a clinicobiological pattern. Methods: In this retrospective study, we assessed 73 adult patients undergoing surgical removal of malignant melanoma. The primary endpoint was histopathologically confirmed lymph-node metastasis. Routinely collected patient-related data, including clinical, anatomical, operative, histopathological, nodal, comorbidity, biological, clinical course, and available staging data, were structured into interpretable constructs. These included the Histopathologic Aggressiveness Index (HAI), the Inflammatory–Metabolic Dysregulation Index (IMDI), the Biological–Histological Discordance Score (BHDS), model-estimated nodal metastatic probability, integrated clinicobiological risk, and explanation stability. The AI-based framework was evaluated by applying bias-reduced and penalized logistic regression, machine learning benchmarking, leave-one-out cross-validation, bootstrap estimation, permutation testing, decision curve analysis, rule extraction, feature stability evaluation, network analysis, similarity-based retrieval, conformal uncertainty estimation, and unsupervised phenomapping. Results: For 72 out of 73 patients, nodal histopathology results were available. Among these patients, 22 had positive nodal status. Positive nodal status was associated with a higher Breslow thickness, an increased mitotic rate, ulceration, lymphovascular invasion, a nodular subtype, and palpable adenopathy. The HAI demonstrated the strongest discriminative signal between node-positive and node-negative patients (median values of 67.8 vs. 45.3; p < 0.001) and retained an independent association with nodal metastasis within the bias-reduced logistic model (odds ratio [OR] per 10-point increase: 2.74; 95% confidence interval [CI]: 1.58–4.75; p < 0.001). The IMDI showed a weak exploratory relationship and did not retain an independent association after adjustment. Similarly, the BHDS did not show significant differences in separating the two endpoint groups. The penalized logistic model including only the HAI showed good performance under leave-one-out cross-validation, with ROC AUC = 0.889, PR-AUC = 0.706, and Brier score = 0.137. Through rule extraction, we found a cohort-specific HAI threshold value > 58.6, above which all node-positive cases were located. With respect to explainability, feature stability, network analysis, similarity retrieval, conformal prediction, and phenomapping, there was convergence toward a dominant high-risk phenotype defined primarily by histopathological criteria. Conclusions: Routine melanoma registries may be transformed into internally evaluated melanoma intelligence frameworks. Histopathologically confirmed lymph-node metastasis among patients with malignant melanoma was organized primarily along an axis of local histopathological aggressiveness, while systemic inflammatory–metabolic dysregulation provided subordinate contextual biological information. Full article

Plant diseases, being a subject of interdisciplinary research, significantly reduce crop yield, quality, and economic returns, while the misidentification of pathogens often leads to ineffective treatments and may harm beneficial organisms and ecosystems. This work develops an approach for robust visual classification of plant diseases under limited and heterogeneous data based on multi-scale fractal texture descriptors integrated into a convolutional neural network. The proposed method employs wavelet transform modulus maxima to extract two complementary fractal characteristics, local fractal dimension and singularity spectrum width, from leaf images at several spatial scales. These descriptors form multi-channel fractal maps fed into a fractal attention module (FAM) inserted after the third stage of a ResNet-50 architecture. The FAM learns to emphasize spatial regions where fractal properties are most discriminative, while a parallel branch encodes global fractal statistics into an auxiliary vector combined with backbone features at the final classification layer. Experiments are conducted on a large heterogeneous collection of 11 public plant disease datasets under 5-shot, 50-shot, and full-scale training regimes. The fractal-augmented model raises classification accuracy from 57.06% to 67.73% on 5 shots and from 80.81% to 86.11% on 50 shots, red outperforming the plain ResNet-50 in these settings, converges within 1–2 epochs versus 25–40, and shows markedly better resilience to color distortions, random occlusions, and grayscale conversion in most cases. The generated attention maps provide spatially explicit explanations of the model’s decisions, increasing transparency for practical use. The proposed approach demonstrates that fractal analysis, embedded as a modulating signal inside a deep network, can serve as an efficient and interpretable inductive bias, which is particularly valuable under data scarcity and noisy agricultural imagery. Full article

Schizophrenia is a highly heterogeneous neuropsychiatric disorder characterized by positive symptoms, negative symptoms, and cognitive impairment, with substantial long-term functional consequences. Although dopaminergic dysfunction remains central to current disease models and treatment, dopamine-centered frameworks alone cannot fully explain cognitive deficits, treatment resistance, or marked variability in therapeutic response. Emerging evidence supports a broader view in which genetic susceptibility and environmental exposures converge on multiple interacting biological systems, including dopaminergic and glutamatergic neurotransmission, neuroimmune and glial dysfunction, kynurenine pathway metabolism, large-scale brain network dysconnectivity, and gut–brain communication. In this review, we integrate these mechanisms within a systems-level framework and discuss how their interactions may contribute to symptom heterogeneity and disease progression. We further examine the limitations of conventional dopamine D2-based antipsychotics, emerging non-dopaminergic pharmacological strategies, and adjunctive interventions targeting cognition and functional recovery. Finally, we highlight major translational barriers, including treatment resistance, medication non-adherence, adverse-effect burden, and the lack of clinically actionable biomarkers. An integrated understanding of converging neurobiological mechanisms may provide a stronger foundation for mechanism-informed patient stratification and precision treatment in schizophrenia. Full article

A delayed malware dissemination model for wireless sensor networks is extended by replacing the single fixed anti-virus cycle time with distributed recovery memories. Two normalized gamma kernels are considered: a one-stage exponential memory (weak kernel) and a two-stage gamma memory (strong kernel), both representing heterogeneous anti-virus removal times across carrier and infectious nodes. The linear chain trick transforms the corresponding integro-differential systems into six- and eight-dimensional ordinary differential equation models. For these distributed-delay formulations, we derive the common disease-free equilibrium and its local stability conditions, prove that the endemic equilibrium and the basic reproduction number are preserved, and show that the inherited delayed-removal mechanism does not generate a positively invariant nonnegative orthant; in its place, we establish the strongest valid conditional boundedness and continuation result. Local endemic stability is reduced to a Routh–Hurwitz problem, and Hopf thresholds are obtained from simple purely imaginary roots together with transversality. For the reference parameter set, full spectral verification and first-Lyapunov-coefficient evaluation show that the first weak- and strong-kernel Hopf points are subcritical. Numerical simulations illustrate convergence below threshold and loss of equilibrium stability above it, while comparison with the discrete-delay benchmark shows that a broader recovery-time distribution postpones instability. Full article

Diabetic retinopathy, diabetic kidney disease, and diabetic peripheral neuropathy remain major causes of visual loss, kidney failure, pain, disability, and reduced quality of life despite improvements in glycemic and cardiovascular risk management. Oxidative stress, mitochondrial dysfunction, iron dyshomeostasis, lipid peroxidation, and sterile inflammation are shared features of these complications and converge on regulated cell-death programs. Ferroptosis is driven by iron-dependent phospholipid peroxidation when glutathione peroxidase 4 and complementary antioxidant systems are insufficient, whereas pyroptosis is an inflammatory lytic process executed by gasdermins after activation of inflammasome-associated or other inflammatory caspases. Nuclear factor erythroid 2-related factor 2 (NRF2) connects these pathways by regulating glutathione synthesis, lipid peroxide detoxification, iron handling, mitochondrial homeostasis, and redox-sensitive inflammatory signaling. Dietary polyphenols may influence this network through electrophilic or kinase-dependent NRF2 activation, preservation of the SLC7A11-glutathione-GPX4 axis, modulation of iron and lipid metabolism, and inhibition of NF-kappaB, TXNIP, NLRP3, caspase-1, and gasdermin signaling. This integrative review critically examines mechanistic, preclinical, and human evidence for these effects in the diabetic retina, kidney, and peripheral nerve. The strongest direct preclinical evidence currently concerns corilagin, resveratrol, isoquercetin, quercetin, epigallocatechin gallate, punicalagin, and selected anthocyanin-rich or phenolic extracts. Human studies suggest possible benefits for albuminuria, retinal edema, endothelial function, and neuropathic outcomes, but they rarely measure ferroptosis- or pyroptosis-specific biomarkers and their results are heterogeneous. The proposed ferroptosis–pyroptosis–NRF2 network should therefore be viewed as a biologically plausible integrative framework rather than a clinically validated linear pathway. Future trials require chemically characterized interventions, exposure biomarkers, tissue-relevant pharmacokinetics, prespecified regulated-cell-death panels, and clinically meaningful microvascular endpoints. Full article

Research on digital banking has largely examined technology adoption, while less is known about how perceived digital capability enters the prior decision of selecting a banking institution. This study adopts a system-based modelling perspective to examine bank selection as a non-linear decision system in which digitalization acceptance, perceived technology level, usage behaviour, and financial education interact to influence the perceived importance of digitalization in bank choice. Using survey data collected from Romanian university students, we develop and evaluate multiple Artificial Neural Network (ANN) configurations to model these relationships and identify the most effective predictive architecture. University students were selected because they represent a highly digitally exposed consumer segment whose banking choices are particularly relevant for examining technology-oriented decision patterns. The results indicate that the optimal ANN configuration achieves strong predictive performance (MAE = 0.1487; MSE = 0.0314) while maintaining efficient convergence behaviour. Variable importance analysis, computed using both a Random Forest measure and permutation importance on the ANN itself, consistently identifies perceived technology level as the dominant component of the decision system and frequency of use as the least influential, while the relative ranking of digitalization acceptance and financial education varies between the two importance measures. The findings demonstrate the suitability of ANN-based approaches for modelling complex behavioural decision systems characterized by non-linear interactions and heterogeneous component influence. The study contributes to the digital banking literature by extending technology acceptance research from adoption intentions toward bank selection behaviour. From a practical perspective, the findings suggest that banks targeting digitally oriented young consumers should prioritize perceived technological capability and the quality of the digital banking experience, including usability, trust, transparency, and usefulness. The comparatively lower predictive of usage frequency and financial education further indicates that strategies focused on strengthening perceived digital value may be more relevant to bank selection than those aimed primarily at increasing routine service usage. Full article

Reservoir impoundment can reactivate pre-existing landslides and reorganize slope topography, thereby changing seepage conditions and subsequent deformation. However, crack mapping, geomorphic interpretation, and hydrodynamic diagnosis are still often treated as separate tasks. This study investigates the Hemenkou (HMK) landslide in the Wudongde Reservoir area, China, using multi-scale space–air–ground observations, including multi-temporal optical satellite images, unmanned aerial vehicle (UAV) photogrammetry, pyramid scene parsing network (PSPNet)-based crack segmentation, global navigation satellite system (GNSS) monitoring, and convergent cross mapping (CCM). The remote sensing record shows a progressive damage sequence: cracks were mainly restricted to the upper source area in 2012, crown cracking intensified and propagated downslope by December 2020, and the UAV survey of 10 June 2024 revealed a mature tension-crack network concentrated in Zone II. ResNet-50-PSPNet achieved the best crack-extraction performance among the tested models, with Precision = 0.9120, Recall = 0.9041, F1 = 0.9081, and IoU = 0.8316. The mapped cracks are dominated by short, narrow, northeast–southwest-oriented tension cracks. GNSS monitoring reveals strong spatial heterogeneity, with stepwise deformation concentrated in Zone II. CCM provides strong directional evidence for the influence of reservoir water-level fluctuation on Zone II deformation, whereas the weaker rainfall signal is consistent with a secondary reinforcing role. The apparent increase in the rainfall-related CCM signal from 2021 to 2023 is consistent with progressive crack expansion and potentially enhanced hydraulic connectivity in Zone II. Taken together, these observations support the interpretation that post-deformation topography, particularly the tension-crack network and disturbed toe, may organise preferential seepage pathways and increase the sensitivity of the landslide to reservoir drawdown. The study provides an integrated remote sensing and monitoring framework for process-based interpretation of reservoir landslides. Full article

New energy power systems with high penetration rates feature complex spatiotemporal coupling relationships among generation, transmission, load, and storage. The random fluctuations in wind and solar power output, combined with load uncertainty, exacerbate system operational risks. Traditional static modeling, single-metric evaluation, and centralized analysis methods struggle to adapt to the dynamic, highly uncertain, and multi-constrained operational scenarios of new power systems. To address this, this paper proposes an Artificial Intelligence-based Comprehensive Evaluation Method for Power System Flexibility and Sustainability (AI-FSEA) under privacy and security constraints. This method first establishes an intelligent fusion module for multi-source, heterogeneous power data, which accurately extracts the system’s multidimensional dynamic features through adaptive wavelet denoising and a temporal self-attention mechanism. Second, it establishes a five-objective coupled evaluation model that balances technical, economic, low-carbon, and reliability considerations, with regulation margin loss, response delay, operating costs, carbon emissions, and power supply instability rate as the core optimization objectives, thereby achieving multi-objective trade-off optimization within the system’s feasible domain; furthermore, a Hierarchical Deep Q-Network-Assisted Multi-Objective Evolutionary Algorithm (HDQN-MOEA) is designed, which leverages the value iteration, composite reward mechanism, and feedback clustering screening mechanism of the deep Q-network to enhance the model’s solution accuracy and convergence efficiency. Results from multiple sets of comparative experiments, ablation studies, and uncertainty generalization experiments conducted using the IEEE standard node system and real-world power grid data from East China indicate that, compared with mainstream optimization algorithms such as NSGA-III and TS-NSGA-II, the proposed HDQN-MOEA algorithm achieves an average improvement of 10.2% in the hypervolume metric and an average reduction of 35.6% in the span metric; the results of the ablation experiments confirm that the absence of the multi-source data fusion module, the hierarchical strategy module, or the feedback clustering screening module would result in a 15.3% and 12.1% decrease in the model’s hypervolume metric, respectively, as well as a slight deterioration in population diversity; under three types of highly uncertain operating conditions—random fluctuations in renewable energy, sudden load spikes, and extreme weather—the algorithm proposed in this paper consistently maintains stable optimization performance, meeting convergence accuracy requirements in as few as 5000 iterations. Without increasing the complexity of existing algorithms, it achieves the synergistic optimization of data privacy and security, evaluation accuracy, and operational efficiency. The proposed method can accurately quantify the dynamic flexibility and long-term sustainability of the new power system, providing reliable intelligent technical support for the planning and dispatch of the new power system, the optimal allocation of resources, and low-carbon, sustainable operation. Full article

Ensuring reliable and efficient communication under high-delay conditions remains a major challenge for Network Digital Twin (NDT) systems operating across heterogeneous networks. This paper experimentally evaluates the Transmission Control Protocol (TCP), User Datagram Protocol (UDP), QUIC, Stream Control Transmission Protocol (SCTP), the DTN7 Bundle Protocol Version 7 (BPv7) implementation, and the NASA Interplanetary Overlay Network (ION) using a controlled delay-generation environment. Performance is assessed through throughput, bandwidth utilization, transmission time, transfer time, End-to-End Completion Time (ECT), and packet inter-arrival variability. The results show that conventional feedback-driven transport protocols suffer significant performance degradation as network delay increases, whereas UDP maintains high throughput but does not guarantee reliable delivery. In contrast, BPv7-based communication mechanisms, particularly ION using the Licklider Transmission Protocol convergence layer (ION-LTP), achieve superior end-to-end responsiveness under the evaluated long-delay conditions. To enable unified comparison across heterogeneous protocol families, this paper introduces ECT as a protocol-independent application-level metric. The experimental benchmark demonstrates that no single communication mechanism is universally optimal, highlighting the need for adaptive communication strategies in NDT systems. Based on the observed protocol trade-offs, a conceptual Artificial Intelligence (AI)-assisted Deep SARSA framework is presented as a potential approach for dynamic protocol selection under varying network conditions. Its implementation, agent training, and experimental validation are reserved for future work. Full article

Pancreatic ductal adenocarcinoma (PDAC) is characterized by extensive adaptive signaling plasticity and metabolic reprogramming that contribute to therapeutic resistance. While the PI3K–mTOR pathway is a central regulator of these processes, the role of the FGF19–FGFR4 axis and its interaction with PI3K-mTOR signaling remains incompletely defined in PDAC. This study investigated whether co-targeting FGFR4 and PI3K-mTOR signaling could overcome adaptive pathway reactivation and suppress tumor-promoting phenotypes. PDAC cell lines representing a spectrum of FGFR4 dependence were treated with the selective FGFR4 inhibitor fisogatinib in combination with the clinically relevant PI3K–mTOR inhibitor gedatolisib. Transcriptomic analyses of TCGA data, along with molecular and functional responses, were evaluated. Transcriptomic analysis demonstrated a positive association between FGFR4 and PI3K–mTOR signaling and linked combined pathway components with poorer overall survival. Across heterogeneous PDAC models, combined inhibition reduced viability, clonogenic survival, migration, and cell-cycle progression more consistently than monotherapy. Apoptosis induction was driven principally by fisogatinib and combination treatment, resulting in comparable apoptotic responses across cell lines. Mechanistically, combined inhibition converged on increased 4E-BP1 inhibitory activity, despite compensatory ERK-RSK pathway activation, indicating that adaptive MAPK was insufficient to restore downstream translational output. In FGFR4-dependent cells, the combined inhibition also reduced secretion of the FGFR4 ligand FGF19. The FGFR4 and PI3K–mTOR signaling comprises a partially interconnected network in PDAC that converges on 4E-BP1-dependent translational control. Dual pathway inhibition consistently and additively suppressed tumor-associated phenotypes despite compensatory MAPK activation, with the clearest added benefit over fisogatinib alone in migration, cell-cycle control, and 4E-BP1 modulation, supporting translational regulation as a shared therapeutic vulnerability. These findings provide a preliminary rationale for biomarker-guided strategies targeting FGFR4 and PI3K–mTOR signaling in pancreatic cancer. Full article