Search Results (353)

Background: Digital marketing increasingly functions as a complex adaptive system characterised by non-stationary environments, strategic interaction, and multi-agent competition. Programmatic advertising exemplifies this complexity, where decisions must be made in real time under uncertainty. Under such conditions, traditional static optimisation methods often fail to deliver robust performance. This review synthesises bio-inspired computational approaches, reinforcement learning (RL), multi-agent reinforcement learning (MARL), and agentic artificial intelligence (AI) to develop an integrated theoretical perspective on adaptive optimisation in digital marketing. Methods: Following PRISMA 2020 guidelines, we conducted a systematic search of peer-reviewed research across six databases: Scopus, IEEE Xplore, ACM Digital Library, SpringerLink, ScienceDirect, and arXiv, supplemented by manual reference checking. Each computational paradigm is explicitly grounded in foundational biological literature, including work on evolution, foraging, swarm intelligence, and immune cognition. Reinforcement learning supports adaptive decision-making through mechanisms closely aligned with operant conditioning and foraging behaviour. Multi-agent reinforcement learning extends these principles to interactive marketing ecosystems via decentralised coordination and swarm-based learning. Agentic AI further advances adaptive capability by introducing goal-directed reasoning, memory, and higher-level decision orchestration. Contributions: The review identifies persistent fragmentation across marketing sub-domains and a lack of formal mathematical grounding for widely used bio-inspired analogies. To address these gaps, the study proposes a multi-layer bio-inspired framework and outlines a structured research agenda to guide the development of autonomous digital marketing systems. Full article

Leading-edge tubercles provide a bio-inspired modification for regulating the hydrodynamic performance of marine propellers, but their controllable generation on complex three-dimensional blades remains insufficiently studied. This study develops a parametric modeling method for tubercled leading-edge propellers based on the David Taylor Model Basin (DTMB) 4383 geometry. The method combines B-spline smooth reconstruction with a Gaussian envelope function to control tubercle radial peak position and height. After validation with publicly available open-water experimental data, the computational fluid dynamics (CFD) method is applied to uniform and locally targeted tubercled models. The results show that leading-edge tubercles modify the suction-side low-pressure region and redistribute local blade loading. The radial peak position mainly controls the concentration region of the pressure disturbance, whereas peak height affects its intensity. At the design advance coefficient, moving the target peak from r/R = 0.26 to r/R = 0.92 decreased the thrust coefficient by 0.77% and increased the torque coefficient by 1.75%. For a fixed target position, increasing the maximum amplitude ratio from 0.05 to 0.80 increased the thrust and torque coefficients by 0.25% and 0.88%, respectively. These findings indicate that tubercle design should balance radial position and protrusion height. Full article

Background: Cyclin-dependent kinase 2 (CDK2) is a key regulator of cell cycle progression and an important therapeutic target in cancer treatment. This study aims to identify novel CDK2 inhibitors using an integrated computational approach combining machine learning and structure-based methods. Methods: A computational pipeline was developed incorporating Lipinski’s Rule of Five filtering, machine learning (ML)-based activity prediction, molecular docking, and molecular dynamics simulations (MDs). A dataset of CDK2 inhibitors with IC50 values was retrieved from ChEMBL, and molecular fingerprints were generated using PaDEL. A 5-fold stratified cross-validation approach was applied to train multiple classifiers, with the random forest model showing the best performance. Predicted active compounds from the InterBioScreen database were subjected to docking against CDK2 (PDB ID: 2FVD) using PyRx, followed by 100 ns MDS for stability analysis. Results: The random forest classifier achieved an AUC-ROC of 0.90 and an accuracy of 0.84. A total of 187 compounds were predicted as active. Among these, two compounds, STOCK4S-00019 and STOCK4S-00025, demonstrated docking scores comparable to the co-crystallized reference ligand. Molecular dynamics simulations confirmed stable binding, consistent interaction patterns, and favorable conformational behavior throughout the simulation period. Conclusions: The identified compounds, STOCK4S-00019 (hit1) and STOCK4S-00025 (hit2), show strong potential as CDK2 inhibitors. These findings support their further investigation through experimental validation and highlight the effectiveness of integrated computational approaches in anticancer drug discovery. Full article

Artificial intelligence (AI) is increasingly influencing nanopharmaceutical development by supporting the transition from empirical formulation screening toward predictive, data-driven, and translationally oriented design. Nanocarrier-based therapeutics are governed by nonlinear relationships among material composition, physicochemical attributes, manufacturing parameters, biological identity, pharmacokinetics, toxicity, and therapeutic performance. In this review, we examine how AI can contribute to nanopharmaceutical development from predictive formulation design to clinical translation. We synthesize current applications of machine learning, deep learning, physics-informed modeling, hybrid mechanistic–AI approaches, and automated optimization workflows, with emphasis on critical quality attribute modeling, multi-objective optimization, design of experiments, quality-by-design, process analytical technology, digital twins, and continuous manufacturing. We also discuss applications involving nano–bio interactions, pharmacokinetics, toxicity, immunogenicity, and precision nanomedicine. AI-based approaches can support rational nanocarrier design, identify nonlinear formulation–property relationships, guide optimization, improve process understanding, and integrate heterogeneous experimental, biological, and manufacturing datasets across diverse nanopharmaceutical platforms. These methods are particularly relevant for modeling protein corona formation, cellular uptake, intracellular trafficking, biodistribution, pharmacokinetics, toxicity, immunogenicity, and patient-specific responses. However, translational implementation remains limited by fragmented datasets, inconsistent reporting standards, limited interpretability, insufficient external validation, uncertain predictions, poorly defined applicability domains, and evolving regulatory expectations for adaptive computational models. Overall, AI should be viewed not only as an optimization tool, but also as a translational framework connecting formulation science, biological prediction, manufacturing control, and clinical implementation. Future progress will depend on standardized data infrastructures, explainable and externally validated models, uncertainty quantification, applicability-domain definition, hybrid mechanistic–AI frameworks, regulatory-ready documentation, and clinically relevant case studies. Full article

Background/Objectives: Advancements in neuroscience and machine learning have increasingly enabled brain pattern recognition based on bio-signal measurements, such as electroencephalography (EEG). These developments support next-generation technologies, including brain–computer interfaces (BCIs) and AI-assisted systems. However, volume conduction (VC) effects remain a major source of contamination in EEG recordings, affecting both univariate analyses and functional connectivity estimation. Methods: In this work, we propose a VC mitigation method that explicitly models and suppresses VC components in the observed functional networks. Specifically, the observed functional network is decomposed into a matrix capturing only VC-related components (i.e., components attributed to volume conduction) and a residual matrix, where the residual is regarded as a proxy for a VC-mitigated functional network that better reflects the underlying functional interactions. The VC component matrix is modeled using a decay function parameterized by the inter-electrode distance matrix, capturing the dominant spatial bias induced by VC. To estimate these parameters, we introduce supervised channel importance, quantified as the mutual information between experimental labels and channel signals, as a proxy for task-relevant neural activity. The parameters are optimized such that the unsupervised node importance derived from the VC-mitigated functional network, defined as the average node strength, aligns with the supervised channel importance. Results: Evaluation results using a deep-learning framework demonstrate that, compared with the observed functional network, the VC-mitigated functional network improves classification performance in brain pattern recognition tasks. Full article

Background: This paper presents the development of digital tools created within the BIORADAR European project to improve user access to Life Cycle Assessment (LCA) results from the project’s use cases and to enable users to upload, benchmark and analyze their own data. The work addresses common challenges in circularity and environmental impact assessment, particularly data availability and expert-assisted self-assessment for users such as small- and medium-sized enterprises. Methods: The LCA data for the project use cases is calculated using the Environmental Footprint methodology. Benchmarking compares bio-approach use cases with traditional approaches across three key sectors selected by the BIORADAR project: fertilizer, textile and packaging. These sectors are recognized by the European Commission as three of the most important sectors in terms of environmental impact. Case impact factor data are normalized using a reference statistic, and a weighting is assigned to each key performance indicator to calculate the global score. Individual impact factor values can also be used for benchmarking. Synthetic data are generated through an advanced statistical decomposition algorithm. Advanced data analytics are provided with clustering and a decision tree algorithm using supervised machine learning. Results: Two examples of decision-oriented case studies are used to illustrate how the platform can support the interpretation and use of already computed LCA results in realistic settings. The web-based expert-assisted self-assessment tool, developed in JavaScript, allows users to input their data, benchmark them against project results and perform multidimensional data analysis. The resulting digital tools provide access to LCA data for each use case, generate realistic synthetic datasets preserving key statistical properties, support benchmarking of both project and user-uploaded cases, and perform data analytics, which complement the benchmarking module with a structural and exploratory interpretation of the data. Conclusions: Overall, the tools integrate use case benchmarking, data processing, advanced analytics and user interfaces to facilitate environmental self-assessment and comparison within the BIORADAR framework. Full article

Background: Biodegradable implants manufactured from Mg-based alloys are one of the most commonly used in orthopedics. However, their overall clinical acceptance is influenced by their fast corrosion speed and hydrogen emission. Based on an innovative manufacturing route previously described, this study introduces a preliminary proof-of-concept for a Gemini-assisted Digital Twin (Gemini-DT),which is an AI-augmented in silico framework designed to consider a MgF₂ conversion coating on the implant surface and to model the synchronization of the degradation process with new bone formation. Methods: Based on the integration of experimental data for Mg-Nd and Mg-Zn alloys and by considering the implant geometry and coating formation, we developed, in collaborative work with LLM Gemini 1.5 Flash (Google), a four-module cognitive framework (surface thermodynamic synergy (Module 1), degradation analysis and alloy extract concentration management (Module 2), micro-channel fluidics and mechanical stability (Module 3), and bio-mechanical synchronization and regenerative evaluation (Module 4)) to evaluate simulated implant behaviors). Results: Using a 10,000 iteration Monte Carlo stability simulation, the model demonstrated a potential 12% reduction in false-negative design screening errors compared to rigid rule-based systems, achieving strong internal decision consistency in sustaining the mandated parametric compliance window. Computational verification supports the projected biocompatibility trends of Mg-Zn alloys, as previously demonstrated in our in vivo studies. Conclusions: Our research leads to a consistent computational architecture dedicated to Mg-based implants and offers a robust platform for virtual design and optimization. These observations suggest that the developed model can recover viable designs, whereas traditional linear models may reject them. Full article

Background: Pathogenic BRCA1 and BRCA2 mutations confer a homologous recombination deficiency that underlies PARP inhibitor sensitivity. While BRCA1 mutation carriers more frequently develop triple-negative breast cancer (TNBC) and BRCA2 carriers hormone receptor-positive (HR+) disease, whether the specific protein domain harboring a pathogenic somatic mutation differs systematically between breast cancer subtypes remains uncertain. Apparent domain enrichment in earlier unfiltered analyses may be confounded by missense variants of uncertain significance (VUSs), which lack clinical actionability. Methods: We assembled three independent breast cancer cohorts via cBioPortal: TCGA-BRCA (brca_tcga_pub2015), METABRIC (brca_metabric), and MSK-CHORD (msk_chord_2024). All somatic BRCA1/2 mutations were mapped to UniProt-annotated functional domains and to Rebbeck-defined breast/ovarian cancer cluster regions (BCCR/OCCR). Per ENIGMA/ACMG guidance, pathogenic mutations (nonsense, frameshift, and canonical splice site) were analyzed inferentially, while missense and in-frame variants—predominantly VUSs—were only reported descriptively. Fisher’s exact tests with Benjamini–Hochberg FDR correction were applied across domain × subtype contingencies. Cohort heterogeneity was assessed via Cochran’s Q and I² statistics; pooled effect estimates were computed using inverse-variance fixed-effects meta-analysis. Results: A total of 394 somatic BRCA1/2 mutations were identified across the three cohorts (BRCA1 n = 166; BRCA2 n = 228), of which 147 (37.3%) met pathogenic criteria. Among 131 pathogenic mutations in HR+/HER2− or TNBC subtypes, 84 (64.1%) occurred in HR+/HER2− disease and 47 (35.9%) in TNBC. Domain-level distributions did not differ significantly between subtypes for any BRCA1 domain (BRCT: TNBC 20.0% vs. HR+ 18.8%, OR = 1.08, 95% CI 0.31–3.78, and FDR-adjusted p = 1.00) or BRCA2 domain (DBD: TNBC 17.6% vs. HR+ 30.8%, OR = 0.48, and FDR-adjusted p = 1.00). Cluster-region analyses (nine Rebbeck BCCR/OCCRs) similarly showed no significant enrichment. Post hoc power analysis indicated that the study could only reliably detect large effects (OR ≥ ~3.0 for the principal BRCT contrast), and formal equivalence testing (TOST) demonstrated equivalence within a prespecified ±20% margin for BRCA1 BRCT (TOST p = 0.031). Heterogeneity across cohorts was minimal (Cochran’s Q = 0.62, I² = 0.0%). Descriptive analyses of VUSs suggested the apparent enrichment of BRCA1 BRCT-localized missense variants in TNBC (31.8% vs. 17.9% in HR+), but this signal did not extend to pathogenic mutations. Conclusions: Within the statistical power available, our three-cohort analysis shows no evidence of large subtype-specific enrichment of pathogenic BRCA1/2 somatic mutations across protein domains or cluster regions; small to moderate effects cannot be excluded. Notably, the majority (64%) of pathogenic mutations occurred in HR+/HER2− disease, underscoring that BRCA1/2 testing should not be deprioritized in non-TNBC subtypes. The apparent BRCT enrichment observed in earlier unfiltered analyses appears to be driven by VUSs rather than pathogenic variants, highlighting the methodological necessity of pathogenicity filtering for clinically actionable inference. These findings provide cohort-scale supportive evidence for emerging clinical guidelines that recommend broader BRCA1/2 testing across breast cancer subtypes. Full article

Flapping-wing flight offers a promising solution for aerial mobility in low-density environments such as the Martian atmosphere, where conventional rotorcraft faces significant performance constraints. However, the coupled aerodynamic and structural mechanisms governing lift generation at low Reynolds numbers remain insufficiently understood. This study investigates the aeroelastic and unsteady aerodynamic behaviour of a bio-inspired flapping wing using an integrated experimental–numerical framework. High-speed imaging is employed to extract representative wing kinematics, including flapping frequency, stroke amplitude, and rotational motion. A geometrically scaled wing model is developed based on Reynolds number similitude and analysed using finite element methods to characterise its dynamic response. Aeroelastic behaviour is evaluated through modal transient simulations, while aerodynamic performance is assessed using both vortex-lattice modelling and computational fluid dynamics. The results show strong coupling between bending and torsional modes, with the structural response highly dependent on excitation frequency relative to the natural modes. Near-resonant conditions lead to amplified deformation and distinct phase relationships, while aerodynamic simulations reveal vortex-dominated lift generation. These findings provide a physics-based framework for the design and analysis of flapping-wing systems operating in low-Reynolds-number and low-density flight regimes. Full article

Agricultural cold chain logistics is characterized by inherent challenges—product perishability, high carbon emissions, and stringent time windows—which are further exacerbated by dynamic disruptions. Existing methods suffer from slow adaptability, unstable multi-objective convergence, and severe cold-start issues. This work falls within the broad scope of biomimetics—the science of emulating nature’s time-tested strategies to solve complex engineering problems—and bio-inspired data-driven methods and their applications in engineering control, optimization, and artificial intelligence. The proposed H-MODRL framework embodies core biomimetic principles: the Genetic Algorithm (GA) mimics Darwinian natural selection and genetic inheritance, the Sparrow Search Algorithm (SSA) abstracts the cooperative foraging and anti-predation behaviors of sparrow populations in nature, and the Arrhenius-based freshness-decay model captures the biochemical kinetics governing perishable biological products. By synergistically integrating these biological evolution principles, swarm intelligence, and deep learning, the framework tackles real-world logistics complexity in a manner directly inspired by living systems. This study presents a well-organized hybrid optimization framework (H-MODRL) that couples a three-stage hybrid evolutionary mechanism, synergistically integrating heuristic warm-start, evolutionary policy guidance, and deep reinforcement learning decision-making. First, an improved genetic algorithm combined with the earliest deadline first strategy constructs a feasible initial population satisfying hard time-window constraints. Second, a large neighborhood search-enhanced chaotic sparrow search algorithm builds a high-quality elite guidance set for policy learning. Third, a physics-based multi-objective proximal policy optimization model embedded with Arrhenius equation-derived freshness-decay kinetics performs online decision-making. Experiments demonstrate that pre-computed all-pairs shortest paths and an O(1) hash-based dynamic-disruption indexing mechanism support fast online replanning. On heterogeneous simulated terrains based on real Chinese geospatial data, H-MODRL outperforms state-of-the-art algorithms across four objectives—logistics cost, carbon emissions, terminal freshness, and delivery time—while exhibiting compact, low-variance performance distributions, thereby validating its engineering robustness and practical value in complex agricultural cold chain environments. Full article

Objectives: Cone-beam computed tomography (CBCT) is central to three-dimensional assessment in oral surgery and implant dentistry; however, CBCT-derived gray values expressed as HU-like units are not equivalent to true CT-derived Hounsfield Units (HU). This brief methodological secondary analysis evaluated the reliability and practical limitations of such values in assessing radiographic changes after peri-implantitis surgery. Methods: The analysis used the imaging protocol and group-level radiological data from a previously published prospective clinical cohort, conducted under the same protocol and ethical approval of the Institutional Ethics Committee of the Medical University of Lublin (KE-0254/248/11/2023; 23 November 2023). The source cohort included 57 patients treated after implant removal for severe peri-implantitis with small-particle dentin (n = 22), Bio-Oss (n = 15), or spontaneous healing without grafting (n = 20). CBCT scans were analyzed in OnDemand3D (version 1.0.11.1007) using manually selected square regions of interest (ROI; 30 × 30 pixels). No external phantom calibration, cross-device normalization, or formal intra-/inter-observer reproducibility assessment was available in the secondary dataset. Results: The previously reported mean study-site values were 779.62 ± 325.92 gray-value units for small-particle dentin, 910.51 ± 155.03 gray-value units for Bio-Oss, and 206.04 ± 174.21 gray-value units for controls. These findings are presented as protocol-dependent attenuation patterns, not as direct material rankings, bone-density thresholds, or proof of regeneration. Variability remained substantial, with study-site coefficients of variation of 41.8%, 17.0%, and 84.6%, respectively, and high adjacent-site variability. Interpretation was constrained by manual ROI placement, lack of calibration, absence of observer-agreement metrics, unequal follow-up timing, and CBCT sensitivity to scatter, beam hardening, field of view, reconstruction settings, and metal-related artifacts. Conclusions: CBCT-derived gray values may be useful as relative indicators of local radiographic attenuation change within a standardized protocol, but they should not be interpreted as absolute measures of bone density. Future regenerative oral surgery studies should combine standardized acquisition, explicit ROI methodology, repeated measurements, observer-agreement analysis, and complementary clinical, radiographic, or histological outcomes. Full article

Recognizing human actions from static images in complex industrial environments remains challenging due to insufficient feature representation and high computational complexity. This issue is particularly critical in power-grid safety monitoring, where improper worker postures (e.g., bending, climbing, falling) can lead to severe accidents and personal injuries, necessitating automated monitoring systems that operate reliably on resource-constrained edge devices. This study proposes a bio-inspired lightweight recognition framework that integrates an improved YOLO-Pose model with a gated recurrent unit (GRU) network. The scientific motivation is grounded in the observation that the human musculoskeletal system achieves highly efficient motion perception through three key mechanisms: hierarchical muscle coordination providing intrinsic rotation invariance, proprioceptive feedback enabling real-time error correction, and selective neural gating reducing redundant information transmission. These biological principles directly inspire our technical contributions: polar-coordinate encoding provides rotation invariance, three-stage filtering mimics proprioceptive feedback, and GRU gating mirrors selective information propagation. Unlike prior approaches that treat pose-based action recognition as a generic computer vision problem, this work explicitly incorporates anatomical structural constraints into the computational pipeline. The framework addresses three research gaps: (1) existing methods lack biomechanically derived invariance properties; (2) GCN-based approaches use fixed topologies that fail to adapt to occlusion patterns; (3) the trade-off between model complexity and accuracy remains unsatisfactory for edge deployment. Experiments on the self-constructed SKPose dataset demonstrate that the proposed method achieves 95.04% accuracy, outperforming ST-GCN by 3.67 percentage points and 2s-AGCN by 1.94 percentage points, with an inference speed of 48 FPS on 8.7 M parameters in underground power-grid environments and provides practical support for biomimetic perception systems and industrial safety monitoring. Full article

The rapid growth of digital communication and large-scale data exchange has increased the demand for advanced cryptographic techniques capable of resisting emerging computational threats. Conventional encryption methods primarily rely on mathematical complexity, which may become vulnerable with the advancement of high-performance computing and future quantum technologies. Biological molecules such as deoxyribonucleic acid (DNA) and RiboNucleic Acid (RNA) provide unique properties, including extremely high storage density, massive parallelism, and complex nucleotide structures that can inspire novel cryptographic mechanisms. This study proposes a bio-inspired cryptographic framework that integrates DNA encoding and RNA-based transformations to enhance data security. In the proposed framework, digital information is first converted into binary format and mapped to nucleotide sequences using a predefined encoding scheme. The encryption process incorporates multiple molecular transformations, including complementary base pairing, sequence permutation, and transcription-inspired DNA-to-RNA conversion to generate a highly randomized ciphertext. Decryption reverses these transformations to reconstruct the original plaintext. Security evaluation demonstrates that the proposed framework produces high entropy outputs, a substantially large key space, and enhanced resistance to statistical and brute-force attacks. The results indicate that DNA and RNA-inspired cryptographic systems can substantially enhance encryption complexity while maintaining reliable data recovery. This research highlights the potential of molecular cryptography as a promising interdisciplinary approach for future secure communication and biological data storage systems. Full article

Objective: Continuous pressure-flow cerebral bio-signals are critical for monitoring cerebrovascular dynamics but are often disrupted by missing data segments caused by artifacts from a variety of sources. These missing segments refer to segments of the signal that do not contain any valid physiological data. Such interruptions fragment the signals, resulting in discontinuities that compromise their overall integrity. Therefore, reconstructing missing values and preserving signal continuity are essential for ensuring the stable computation of signal trajectories and the accuracy of derived cerebrovascular indices. Methods: To address this issue, this systematic scoping review aimed to identify and synthesize existing interpolation and imputation strategies for handling missing segments in continuous pressure-flow cerebral bio-signals. Following the Cochrane and Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines, a comprehensive search of five electronic databases was conducted from their inception to 24 September 2024, and updated on 16 June 2025, using a detailed search string. Results: The initial searches yielded 19,403 results, and 8 studies were filtered and included in the review. All included studies employed interpolation techniques, such as linear and spline interpolation algorithms, to correct distorted signal segments. However, none of the included studies directly utilized interpolation or imputation strategies to reconstruct or completely fill missing data segments. Conclusions: This reveals a critical knowledge gap, as no study has systematically addressed the utilization of interpolation or imputation strategies for missing segments in pressure-flow cerebral bio-signals. Therefore, this systematic review emphasizes the need for specialized methodologies and standardized frameworks to enable reliable recovery of missing data segments in pressure-flow cerebral bio-signals, which is critical for advancing real-time neurocritical care monitoring and experimental neuroscience/psychological research. Significance: This systematic review lays the groundwork for future research into physiologically informed interpolation and imputation strategies for pressure-flow cerebral bio-signals in clinical and research applications. Full article

Polymer-based product use is rapidly increasing worldwide, resulting in critical social, environmental, ecological, economic, and health effects. Worldwide efforts have increasingly focused on solutions to the equilibrium consumption, production, and disposal of plastics to tackle these issues. The frontiers of biodegradable and bio-based polymers are continually advancing in pursuit of sustainability. Therefore, designing ecological bioplastics made of both biodegradable and bio-based polymers reveals chances to overcome plastic pollution and resource depletion. Polymeric materials are mainly used to manufacture different products at the beginning of their lifespans and which become waste after usage. Numerous sustainability strategies and polymer recycling methods are described and mostly classified into chemical, mechanical, and thermal recycling processes. This manuscript presents a New Polymers Frontier in Recycling and Sustainability Using an Ensemble of Deep Learning with a Heuristic Search Algorithm (NPFRS-EDLHSA). This work is devoted to computational polymer typology, which is based on machine learning algorithms applied to data on physicochemical properties. Although polymer classification can facilitate downstream materials research, the present study does not directly simulate recycling, environmental impacts, or sustainability. The main contributions made by this work include (i) an exploratory analysis of ensemble deep learning models to classify polymers by type on a small and unbalanced dataset; (ii) an evaluation of the effect of feature selection with a heuristic optimization methodology; and (iii) a comparison of the effects on classification performance under limited data conditions. This research sets out to provide a methodological explanation, not arguments for industrial-scale applicability. For the polymer-type classification process, the proposed NPFRS-EDLHSA model designs an ensemble of deep learning techniques, namely a bidirectional recurrent neural network (BiRNN) model, a bidirectional gated recurrent unit (BiGRU) method, and a graph autoencoder (GAE) technique. Finally, the grasshopper optimization algorithm (GOA) adjusts the hyperparameter values of the ensemble models optimally and results in an improved classification performance. A wide-ranging set of experiments was conducted to validate the performance of the NPFRS-EDLHSA method. The experimental results indicated that the NPFRS-EDLHSA technique achieved a better performance than an existing model. Full article