Knowledge-based Approach Research Articles

Mean-Field Coupling Between Local Interactions in Proteins in Relation to Chirality, Secondary, and Supersecondary Structure Formation, and Allostery.

Coarse graining is usually considered as a tool to extend the time and size scale of simulations. However, leaving out the atomistic details to keep their fingerprints in a coarse-grained model also enables us to understand better structure formation and dynamics. In this chapter, by using our scale-consistent theory of coarse graining, we demonstrate that the coarse-grained terms corresponding to the coupling between local conformational states of amino-acid residues explain secondary-structure propagation along polypeptide backbone to stabilize -helices and -strands in proteins and direct the loops preceding and following such segments of protein structure. These and related correlations are probably the still missing terms in both physics- and knowledge-based approaches to protein-structure modeling, including AlphaFold. We also show that the chirality of coarse-grained torsional potentials and, thereby, that of polypeptide backbone emerge from putting together achiral site-based torsional potentials given the phase shift due to residue chirality, and that the improper-torsional potentials that correspond to the coupling between local conformational states of the sites adjacent to a given -carbon atom enable us to model amino-acid-residue enantiomerization.

Enhancing Drug Recommender System for Peptic Ulcer Treatment

Drug Recommender Systems (DRS) streamline prescription process and contribute to better healthcare. Hence, this study developed a DRS that recommends appropriate drug(s) for the treatment of an ailment using Peptic Ulcer Disease (PUD) as a case study. Patients’ and drug data were elicited from MIMIC-IV and Drugs.com, respectively. These data were analysed and used in the design of the DRS model, which was based on the hybrid recommendation approach (combining the clustering algorithm, the Collaborative Filtering approach (CF), and the Knowledge-Based Filtering approach (KBF)). The factors that were considered in recommending appropriate drugs were age, gender, body weight, allergies, and drug interactions. The model designed was implemented in Python programming language with the Flask framework for web development and Visual Studio Code as the Integrated Development Environment. The performance of the system was evaluated using Precision, Recall, Accuracy, Root Mean Squared Error (RMSE) and usability test. The evaluation was carried out in two phases. Firstly, the CF component was evaluated by splitting the dataset from MIMIV-IV into a 70% (60,018) training set and a 30% (25,722) test set. This resulted in a precision score of 85.48%, a recall score of 85.58%, and a RMSE score of 0.74. Secondly, the KBF component was evaluated using 30 different cases. The evaluation for this was computed manually by comparing the recommendation results from the system with those of an expert. This resulted in a precision of 77%, a recall of 83%, an accuracy of 81% and an RMSE of 0.24. The results from the usability test showed a high percentage of performance of the system. The addition of the KBF reduced the error rate between actual recommendations and predicted recommendations. So, the system had a high ability to recommend appropriate drug(s) for PUD.

A Knowledge-Based Molecular Single-Source Precursor Approach to Nickel Chalcogenide Precatalysts for Electrocatalytic Water, Alcohol, and Aldehyde Oxidations.

The development and comprehensive understanding of nickel chalcogenides are critical since they constitute a class of efficient electro(pre)catalysts for the oxygen evolution reaction (OER) and value-added organic oxidations. This study introduces a knowledge-based facile approach to analogous NiE (E = S, Se, Te) phases, originating from molecular β-diketiminato [Ni2E2] complexes and their application for OER and organic oxidations. The recorded activity trends for both target reactions follow the order NiSe > NiS > NiTe. Notably, NiSe displayed efficient performance for both OER and the selective oxidation of benzyl alcohol and 5-hydroxymethylfurfural, exhibiting stability in OER for 11 days under industrially pertinent conditions. Comprehensive analysis, including quasi in situ X-ray absorption and Raman spectroscopy, in combination with several ex situ techniques, revealed a material reconstruction process under alkaline OER conditions, involving chalcogen leaching. While NiS and NiSe experienced full chalcogen leaching and reconstruction into NiIII/IV oxyhydroxide active phases with intercalated potassium ions, the transformation of NiTe is incomplete. This study highlights the structure-activity relationship of a whole series of analogous nickel chalcogenides, directly linking material activity to the availability of active sites for catalysis. Such findings hold great promise for the development of efficient electrocatalysts for a wide range of applications, impacting various industrial processes and sustainable energy solutions.

Enhancing Estimation Accuracy of Prostate Cancer VMAT Planning: A Knowledge-based Approach Using Multiple Collimator Angles.

This study aimed to determine whether a knowledge-based planning model incorporating treatment plans with multiple collimator angles (Multi-coll. model) provides superior estimation accuracy and plan quality for a range of collimator angles compared with a typical institutional model (Inst. model). Five institutions using volumetric modulated arc therapy for prostate cancer participated in the study. The Inst. model comprised plans using a fixed collimator angle, whereas the Multi-coll. model registered plans using single arcs and collimator angles of 10°, 30°, 50°, and 70°. The coefficient of determination and mean squared error were calculated, and the estimation accuracy and dosimetric results for three validation cases at each institution were compared for each model. Dosimetric parameters included volumes receiving at least 20%, 50%, and 90% of the prescription dose (V20, V50, and V90, respectively) at each collimator angle. The Multi-coll. model yielded a higher coefficient of determination and lower mean squared error than the Inst. model, although overfitting of data was a concern at two institutions. The mean absolute errors between estimated and calculated values were (Inst. model vs. Multi-coll. model): 4.41% vs. 2.60% (V20), 4.03% vs. 2.27% (V50), and 2.07% vs. 1.09% (V90) for the rectum; 3.48% vs. 3.42% (V20), 2.24% vs. 3.65% (V50), and 1.11% vs. 0.73% (V90) for the bladder. The Multi-coll. model exhibited a lower rectal and vesical V20 than the Inst. The Multi-coll. model had a greater estimation accuracy and slightly higher plan quality than the Inst.

Automated optimisation design of barge structures using integrated knowledge-based approaches

ABSTRACT An integrated automatic optimisation simulation system for barge structure design has been developed to enhance efficiency and quality. This system incorporates a knowledge-based geometrical modelling method for barge structure design, allowing for the automatic construction of computational models of barge sections through direct transfer of geometric models. Furthermore, a barge structure optimisation simulation system integrated with direct calculation has been established. The effectiveness of this approach is demonstrated through the successful application of structural optimisation to a 65 m barge section, resulting in higher quality design solutions. This research offers an efficient and effective method for achieving high-quality design solutions in the field of barge structure design.

Physiological and molecular insights into the effect of a seaweed biostimulant on enhancing fruit yield and drought tolerance in tomato

Tomato is one of the most widely grown vegetable crops in Europe. This study describes an approach for treating tomato plants with an extract of Ascophyllum nodosum (ANF) prior to a stress event, which prepares the plants at the molecular level to respond more effectively to stress conditions, through a process known as molecular priming. Solanum lycopersicum L. cv. Micro-Tom, a dwarf tomato variety, was pre-treated with ANF via foliar spray during the flowering phase and subsequently subjected to drought conditions. ANF-treated plants exhibited enhanced growth, fruit yield, and stress tolerance under both moderate and severe drought conditions compared to untreated plants. Transcriptomic studies in leaves revealed that the priming treatment preserved the photosynthetic machinery, inducing stress-protective genes involved in ascorbate, peroxidase, GABA, glutathione, and flavanol biosynthesis. Simultaneously, the treatment repressed key senescence-related genes associated with ethylene biosynthesis, as well as several WRKY and NAC transcription factors. Metabolome analysis demonstrated that ANF induces drought tolerance by promoting the accumulation of stress-protective primary and secondary metabolites, such as GABA, proline, maltose, ascorbic acid, quercetin, and biotin, which can act as osmoprotectants and free radical scavengers during drought. Combined transcriptome and metabolome analyses suggest that ANF treatment represses the senescence process, maintains photosynthetic activity, and induces the accumulation of protective metabolites and amino acids, promoting plant survival and growth under drought. Overall, this research provides new insights into the molecular mechanisms underlying biostimulant-based molecular priming and offers a knowledge-based approach for the accurate application of this ANF molecular priming agent to increase crop productivity and mitigate yield loss during drought, contributing to food security in the era of climate change.

Professionalism assessment of students in clinical education through Situational Judgment Test (SJT)

BackgroundThe study aims to assess the situational judgment capability of students in various professions, including medicine, surgical nursing, anesthesia nursing, and emergency medical technology, using a validated and adapted Situational Judgment Test (SJT).MethodsThe cross-sectional study was conducted at Qom University of Medical Sciences in 2023–2024. The study consisted of two steps: (1) adaptation and validity assessment of the SJT in various health professions, and (2) evaluation of students’ situational judgment capability using the adapted SJT. Participants included 207 students from surgical nursing, anesthesia nursing, medicine, and emergency medical technology departments, selected through stratified random sampling. The SJT included 10 constructed-response scenarios developed by Alkhuzaee and colleagues in 2022. The validation and adaptation process utilized consensus methods, and the SJT was modified to a selected response format based on a knowledge-based approach. Scores ranged from 0 to 30. Cohen’s method and norm-referencing standard deviation were used to calculate a cut-off score. Data were analyzed using descriptive statistics (frequency, mean, standard deviation) and inferential statistics (independent t-test, ANOVA, and ANCOVA).ResultsThe face and content validity of the adapted SJT with a selected-response format was confirmed by expert agreement. The reliability of the SJT was approved, with a Cronbach’s Alpha of 0.81. The mean score of students’ situational judgment in professionalism was 11.90 (SD 4.54). Students’ professions significantly impacted their situational judgment scores (F (3.201) = 3.67, p = 0.01, Partial Eta Squared = 0.052).ConclusionThe adaption and validation of SJT with a selected response format were confirmed by experts who evaluated the situational judgment capabilities of students in different clinical professions. The reliability of the SJT was acceptable—Cohen’s method as a standard-setting approach assisted in identifying students needing support and training. The moderate level of situational judgment capacity confirmed the development of the educational program for the improvement of non-cognitive and non-technical skills such as judgment and decision-making in the field of professionalism.

Data-driven homogenisation of viscoelastic porous elastomers: Feedforward versus knowledge-based neural networks

A computational framework is established to implement time-dependent data-driven surrogate constitutive models for the homogenised mechanical response of porous elastomers at large strains. The aim is to enhance the computational efficiency of multiscale analyses through the use of these surrogate models. To achieve this, explicit finite element (FE) simulations are conducted to predict the homogenised response of a cubic unit cell of a porous elastomer, using two different viscoelastic descriptions of the parent material, subject to pseudo-random, multiaxial, non-proportional histories of macroscopic strains. The histories of homogenised variables extracted from each set of FE predictions form a training dataset, which is used to train two different surrogate models, both relying on artificial neural networks (NNs). The first model predicts the increment in macroscopic stress over a simulation step, as a function of the macroscopic stress and strain at the beginning of the step, of the prescribed macroscopic strain increment, and of the corresponding time increment. The second model uses the same inputs and outputs but tests a knowledge-based modelling approach: it relies on the aid of an additional nonlinear elastic constitutive model, which is time- and path-independent and known a priori. This model represents an existing base of knowledge which is augmented and corrected by a NN after training on viscoelastic data. The data-driven surrogate model, therefore, learns the viscoelastic behaviour of the unit cell starting from knowledge of its elastic response. The two surrogate models are found to have comparable and very high accuracies, capturing the response of the homogenised unit cell to complex loading histories. Hyperparameter optimisation shows that the second, knowledge-based model requires a simpler NN and therefore incurs a smaller computational cost.

Knowledge-oriented management systems for complex objects

The article examines the issue of improving the efficiency of complex systems management through the use of knowledge-based methods and technologies. The article emphasizes the need to revise traditional approaches to scientific research and knowledge management by introducing a transdisciplinary paradigm and artificial intelligence to achieve sustainable development and knowledge evolution. The authors try to solve the problem of management in the face of uncertainty and incomplete information, which is typical for such systems. The methods of integrating knowledge from different fields to create an integrated management system are considered. The basis of management is the holarchic structure, which organizes the system in the form of a hierarchy of holons. Each holon has a specific role and knowledge, which ensures the flexibility and adaptability of the system. The study also uses self-organization and learning methods to ensure continuous improvement of the systemʼs functioning. The article discusses the requirements for knowledge-based management systems, such as flexibility, adaptability, data and knowledge integration, learning and self-organization, transparency of decisions, and resilience to errors and failures. The findings of the study emphasize the importance of implementing knowledge-based approaches in the management system of complex facilities. This allows to increase the level of adaptability and stability of management, which is critical in todayʼs environment of dynamic development of technology and knowledge. The proposed methods can be applied in various fields, which makes them universal and useful for a wide range of scientific and practical tasks, providing a high level of adaptability, efficiency and stability of complex objects management, taking into account modern requirements and challenges.

Explaining the Anomaly Detection in Additive Manufacturing via Boosting Models and Frequency Analysis

Anomaly detection is an important feature in modern additive manufacturing (AM) systems to ensure quality of the produced components. Although this topic is well discussed in the literature, current methods rely on black-box approaches, limiting our understanding of why anomalies occur, making complex the root cause identification and the consequent decision support about the action to take to mitigate them. This work addresses these limitations by proposing a structured workflow designed to enhance the explainability of anomaly detection models. Using the wire arc additive manufacturing (WAAM) process as a case study, we examined 14 wall structures printed with INVAR36 alloy under varying process parameters, producing both defect-free and defective parts. These parts were classified based on surface appearance and welding camera images. We collected welding current and voltage data at a 5 kHz sampling rate and extracted features from both time and frequency domains using a knowledge-based approach. Isolation Forest, k-Nearest Neighbor, Artificial Neural Network, XGBoost, and LGBM models were trained on these features, and the results shown best performance of boosting models, achieving F1 scores of 0.927 and 0.945, respectively. These models presented higher performance compared to other models like k-Nearest Neighbor, whereas Isolation Forest and Artificial Neural Network posses lower performance due to overfitting, with an F1 score of 0.507 and 0.56, respectively. Then, by leveraging the feature importance capabilities of these models, we identified key signal characteristics that distinguish between normal and anomalous behavior, improving the explainability of the detection process and in general about the process physics.

Leaders’ perceptions of integrating health and social services after a structural reform in Finland

Abstract Background Health and social service system was reformed in Finland at the beginning of 2023. The responsibility for organizing health and social services was transferred from c. 300 municipalities to 22 well-being services counties (WBSCs). One aim of the reform was to support integration between and within health and social services. The purpose of this study is to describe strategic leadership approaches to implementing integrated care. The study highlights the issues that should be paid attention to when integrated care practices are implemented. Methods The qualitative study utilized semi-structured interviews collected from the upper-level leaders of the WBSCs (n = 25) responsible for the implementation of the reform. The interviewees represented 11 WBSCs. The data were collected in May-August 2023. Inductive content analysis was used for data analysis. Results Four strategic leadership approaches to implementing integration were identified: 1) service user-based leadership approach, 2) partnership-based leadership approach, 3) service system-based leadership approach, and 4) knowledge-based leadership approach. Most of the interviewees combined several different leadership approaches in their talk. Each approach emphasized different aspects of implementing integration and the contextual factors that have to be addressed at different levels of the health and social service system. Conclusions While the macro level reform toward more integrated health and social service system was implemented, the findings reveal that more efforts are required at the meso and micro levels of the system to fulfill the goal of integrated services. The paper highlights the importance of strategic leadership and its different aspects in implementing integrated care at different levels of the system. Key messages • The paper provides a way of comprehending integration from the perspective of strategic leadership. • The conceptualization through the different leadership approaches may facilitate forming a shared vision for integration among leaders and managers.

The quest for the best target genes for RNAi-mediated pest control.

RNA interference (RNAi) has emerged as an eco-friendly alternative to classic pesticides for pest control. This review highlights the importance of identifying the best target genes for RNAi-mediated pest control. We argue that the knowledge-based approach to predicting effective targets is limited by our current gaps of knowledge, making unbiased screening a superior method for discovering the best target processes and genes. We emphasize the recent evidence that suggests targeting conserved basic cellular processes, such as protein degradation and translation, is more effective than targeting the classic pesticide target processes. We support these claims by comparing the efficacy of previously reported RNAi target genes and classic insecticide targets with data from our genome-wide RNAi screen in the red flour beetle, Tribolium castaneum. Finally, we provide practical advice for identifying excellent target genes in other pests, where large-scale RNAi screenings are typically challenging.

Knowledge-based and Expert Systems in Prognostics and Health Management: a Survey

Prognostics and Health Management (PHM) has become increasingly popular in recent years, and data-driven methods and artificial intelligence have emerged as dominant tools within the PHM field. This trend is mainly due to the increasing use of sensors and the ability of machine learning techniques to leverage condition monitoring data. However, despite their utility and effectiveness, these techniques are not without drawbacks. One major issue is that data-driven methods often lack transparency in their reasoning, which is crucial for understanding fault occurrences and diagnostics. Additionally, the availability of data can be a challenge. In some cases, data are scarce or hard to obtain, either due to the cost of installing necessary sensors or the rarity of the required information. Lastly, the insights derived from data can sometimes diverge from those obtained through expert analysis and established norms. This contrasts with knowledge-based approaches such as expert systems, which formally organize the knowledge acquired from norms and experts, and then deduce the desired conclusion. While research is increasingly exploring data-driven techniques, industry tends to still frequently employ knowledge-based methods. To fill this gap, this paper offers a detailed survey of knowledge-based and expert systems in PHM, examining methodologies such as propositional logic, fuzzy logic, Dempster-Shafer theory and Bayesian networks. It assesses the integration and impact of these techniques in PHM for fault detection, diagnosis and prognosis, highlighting their strengths, limitations, and potential future developments. The study provides a thorough evaluation of current developments and contributes significant insights into the current capabilities and future directions of knowledge-based techniques in enhancing decision-making processes in PHM.

Eliminating the Deadwood: A Machine Learning Model for CCS Knowledge-Based Conformational Focusing for Lipids.

Accurate elucidation of gas-phase chemical structures using collision cross section (CCS) values obtained from ion-mobility mass spectrometry benefits from a synergism between experimental and in silico results. We have shown in recent work that for a molecule of modest size with a proscribed conformational space we can successfully capture a conformation(s) that can match experimental CCS values. However, for flexible systems such as fatty acids that have many rotatable bonds and multiple intramolecular London dispersion interactions, it becomes necessary to sample a much greater conformational space. Sampling more conformers, however, accrues significant computational cost downstream in optimization steps involving quantum mechanics. To reduce this computational expense for lipids, we have developed a novel machine learning (ML) model to facilitate conformer filtering according to the estimated gas-phase CCS values. Herein we report that the implementation of our CCS knowledge-based approach for conformational sampling resulted in improved structure prediction agreement with experiment by achieving favorable average CCS prediction errors of ∼2% for lipid systems in both the validation set and the test set. Moreover, most of the gas-phase candidate conformations obtained by using CCS focusing achieved lower energy-minimum geometries than the candidate conformations without focusing. Altogether, the implementation of this ML model into our modeling workflow has proven to be beneficial for both the quality of the results and the turnaround time. Finally, while our approach is limited to lipids, it can be readily extended to other molecules of interest.

Market Orientation, Supply Chain Integration, and Marketing-Technical Integration: Antecedents of Competitive Advantage

The study aims to investigate the role of knowledge-based dynamic capabilities in achieving competitive advantage. First, the study considers the firms’ important marketing and supply chain management capabilities and logically identifies market orientation, supply chain integration, and marketing-technical integration as knowledge-based dynamic capabilities, then empirically attempts to examine their role in achieving competitive advantage. A quantitative research design was utilized, and all constructs were directly measured using multi-item scales. Data was collected from 625 pharmaceutical firms working in Pakistan and analyzed by using partial least squares structural equation modeling. The results show that market orientation has the strongest effect, followed by supply chain integration, and marketing-technical integration. The study provides strong evidence for the need for a knowledge-based approach in firms. Firms must develop knowledge-based dynamic capabilities for acquiring, generating, and combining knowledge to achieve a competitive position in their respective industries. Managers should focus on customer needs and competitor activities, integrate with supply chain partners, and promote a culture of cross-functional collaboration. The study contributes to the existing body of knowledge on knowledge-based dynamic capabilities and emphasizes the importance of a knowledge-based approach in achieving competitive advantage.

How knowledge-based HRM practices affects aspects of innovation capability through knowledge management: the moderating role of innovative culture

Purpose Given the important of knowledge resource and human capital for improving innovation competence, the purpose of this study is to examine the influence of knowledge-based HRM practices on product and process innovation of firms via the mediating role of knowledge management capability and moderating role of innovative culture. Design/methodology/approach The paper used structural equation modeling and empirical data collected from 271 participants in 156 manufacturing and service firms to examine the level of how knowledge-based HRM practices and knowledge management affect product and process innovation under the moderating role of innovative culture. Findings The research findings confirm the mediating roles of knowledge management between knowledge-based HRM practices and two specific types of innovation namely product and process innovation. It also firstly reveals the positive moderating role of innovative culture in enhancing the effects of knowledge management on product innovation. The results underline the necessity of building an innovative climate and knowledge-based HRM practices to stimulate knowledge management for improving innovation capability of firms in the developing and emerging markets. Research limitations/implications The paper helps bring deeper insights to leaders and practitioners about the new knowledge-based approach that enhances innovation competence for organizations. Originality/value The paper significantly contributed to theoretical and practical initiatives on theory of HRM practices and knowledge management by showing different moderating and mediating mechanism thereby firms can follow to enhance innovation capability of firms in developing and emerging markets.

Toward Memory-Efficient Analog Design Using Precomputed Lookup Tables

Analog design productivity remains a challenge in the digitally driven semiconductor chip design field. Knowledge-based and simulation-based analog automation approaches have not achieved widespread acceptance in the analog design community. Systematic analog design using precomputed lookup tables (LUTs) is a promising approach that can address the design productivity challenge. Although modern computing systems have powerful memory capabilities, which make the LUT approach viable, reducing the memory footprint of the LUTs remains a challenge. A memory-efficient design technique using LUTs is proposed by using an incomplete grid in the MOSFET degrees-of-freedom (DoFs) space. An efficient indexing technique for the incomplete grid is also proposed, using a precomputed offset array in various scenarios, such as two-sided constraints and three-dimensional LUTs. The results show that the proposed technique can achieve up to a 67% reduction in memory footprint, in addition to improving LUT generation time and query performance.

Hybrid attribute-based recommender system for personalized e-learning with emphasis on cold start problem

This article introduces a recommendation system that merges a knowledge-based (attribute-based) approach with collaborative filtering, specifically addressing the challenges of the pure-cold start scenario in personalized e-learning. The system generates learning recommendations by assessing item similarities, utilizing the Rogers-Tanimoto similarity measure for materials and users, and Jaccard's similarity for user comparisons. Unlike traditional collaborative methods relying on prior ratings, this approach depends on attributes. Additionally, user and learning material profiling structures were created to serve as fundamental inputs for the recommendation algorithm. These profiles represent student and material knowledge in a two-dimensional space to facilitate matching. Our processes incorporate user learning styles, preferences, and prior knowledge as metrics for achieving the desired level of personalization. The system produces a list of top recommendations based on predicted ratings. To validate its efficacy, a website resembling our learning platform was developed and tested by users. The primary results demonstrate the system's ability to identify similar users even in a pure cold start condition without existing ratings. Consequently, the system proves its capability in recommending suitable materials, modeling students, and identifying similar user groups. The evaluation results of the proposed system showed a good level of satisfaction by the testimonials, quantified by a score of 82% for the recommended materials (16% higher than exiting cold-start systems), and an average score of 90% in terms of satisfaction about the generated student profiles. As they proved the capability of the framework in recommending suitable materials, and its capability in modeling students, finding similar groups of users.

Rapid discovery and evolution of nanosensors containing fluorogenic amino acids

Binding-activated optical sensors are powerful tools for imaging, diagnostics, and biomolecular sensing. However, biosensor discovery is slow and requires tedious steps in rational design, screening, and characterization. Here we report on a platform that streamlines biosensor discovery and unlocks directed nanosensor evolution through genetically encodable fluorogenic amino acids (FgAAs). Building on the classical knowledge-based semisynthetic approach, we engineer ~15 kDa nanosensors that recognize specific proteins, peptides, and small molecules with up to 100-fold fluorescence increases and subsecond kinetics, allowing real-time and wash-free target sensing and live-cell bioimaging. An optimized genetic code expansion chemistry with FgAAs further enables rapid (~3 h) ribosomal nanosensor discovery via the cell-free translation of hundreds of candidates in parallel and directed nanosensor evolution with improved variant-specific sensitivities (up to ~250-fold) for SARS-CoV-2 antigens. Altogether, this platform could accelerate the discovery of fluorogenic nanosensors and pave the way to modify proteins with other non-standard functionalities for diverse applications.

Knowledge-based machine learning for predicting and understanding the androgen receptor (AR)-mediated reproductive toxicity in zebrafish

Traditional methods for identifying endocrine-disrupting chemicals (EDCs) that activate androgen receptors (AR) are costly, time-consuming, and low-throughput. This study developed a knowledge-based deep neural network model (AR-DNN) to predict AR-mediated adverse outcomes on female zebrafish fertility. This model started with chemical fingerprints as the input layer and was implemented through a five-layer virtual AR-induced adverse outcome pathway (AOP). Results indicated that the AR-DNN effectively and accurately screens new reproductive toxicants (AUC = 0.94, accuracy = 0.85), providing potential toxicity pathways. Furthermore, 1477 and 2448 chemicals that could lead to infertility were identified in the plastic additives list (PLASTICMAP, n = 7112) and the Inventory of Existing Chemical Substances in China (IECSC, n = 17741), respectively. Colourants containing steroid-like structures are the major active plastic additives that might lower female zebrafish fertility through AR binding, DNA binding, and transcriptional activation. While active IECSC chemicals primarily have the same fragments, such as benzonitrile, nitrobenzene, and quinolone. The predicted toxicity pathways were consistent with existing fish evidence, demonstrating the model’s applicability. This knowledge-based approach offers a promising computational toxicology strategy for predicting and characterising the endocrine-disrupting effects and toxic mechanisms of organic chemicals, potentially leading to more efficient and cost-effective screening of EDCs.