Relevant Nodes Research Articles

Social networks from dailies: the observer’s point of view

In this paper, we explore the hypothesis that the social network described by an observer in a written document provides a discernible cross-section of the underlying social network. Specifically, we propose that the subnetwork described by the observer is extracted from large social network with a Barabási-Albert structure, following two key premises: (1) the observer identifies and reports nodes and interactions in his local environment, and (2) the observer records the most relevant nodes—those with a high number of connections—as well as the most significant portions of their local environment. We test this hypothesis using personal dailies written in New Spain and Mexico between 1776 and 1873. We compare the structure of these dailies networks to one depicted by a typical subnetwork of a Barabási-Albert network, obtained following the two premises mentioned above. To support our findings, we compare them to the social networks described in two well-known novels. In this work, we implicitly assume the hypothesis that there is a complete, non-subjective social network, including our observer, with a Barabási-Albert structure. The evolution of this underlying social network can shed light on the dynamics of a society over time. Our approach could give place to indicators of the changes in the structure of this enveloping network. This is especially important since, although our historical sources do not offer sufficient data to reconstruct the processes in their entirety, the approach could supply clues or fragments from which we could understand the structure and dynamics of societies over time.

Power Quality State Estimation for Distribution Grids Based on Physics-Aware Neural Networks—Harmonic State Estimation

In the transition from traditional electrical energy generation with mainly linear sources to increasing inverter-based distributed generation, electrical power systems’ power quality requires new monitoring methods. Integrating a high penetration of distributed generation, which is typically located in medium- or low-voltage grids, shifts the monitoring tasks from the transmission to distribution layers. Compared to high-voltage grids, distribution grids feature a higher level of complexity. Monitoring all relevant nodes is operationally infeasible and costly. State estimation methods provide knowledge about unmeasured locations by learning a physical system’s non-linear relationships. This article examines a new flexible, close-to-real-time concept of harmonic state estimation using synchronized measurements processed in a neural network. A physics-aware approach enhances a data-driven model, taking into account the structure of the electrical network. An OpenDSS simulation generates data for model training and validation. Different load profiles for both training and testing were utilized to increase the variance in the data. The results of the presented concept demonstrate high accuracy compared to other methods for harmonic orders 1 to 20.

GASCOM: Graph-based Attentive Semantic Context Modeling for Online Conversation Understanding

Online conversation understanding is an important yet challenging NLP problem which has many useful applications (e.g., hate speech detection). However, online conversations typically unfold over a series of posts and replies to those posts, forming a tree structure within which individual posts may refer to semantic context from elsewhere in the tree. Such semantic cross-referencing makes it difficult to understand a single post by itself; yet considering the entire conversation tree is not only difficult to scale but can also be misleading as a single conversation may have several distinct threads or points, not all of which are relevant to the post being considered. In this paper, we propose a Graph-based Attentive Semantic COntext Modeling (GASCOM) framework for online conversation understanding. Specifically, we design two novel algorithms that utilize both the graph structure of the online conversation as well as the semantic information from individual posts for retrieving relevant context nodes from the whole conversation. We further design a token-level multi-head graph attention mechanism to pay different attentions to different tokens from different selected context utterances for fine-grained conversation context modelling. Using this semantic conversational context, we re-examine two well-studied problems: polarity prediction and hate speech detection. Our proposed framework significantly outperforms state-of-the-art methods on both tasks, improving macro-F1 scores by 4.5% for polarity prediction and by 5% for hate speech detection. The GASCOM context weights also enhance interpretability.

Sparse dynamic graph learning for district heat load forecasting

Accurate heat load forecasting is crucial for the efficient operation and management of district heating systems. This study introduces a novel Sparse Dynamic Graph Neural Network (SDGNN) framework designed to address the complexities of forecasting heat load in district heating networks. The proposed model represents the district heating network as a dynamic graph, with nodes corresponding to consumers or heat sources and edges denoting temporal dependencies. The SDGNN framework comprises three key components: (1) a sparse graph learning module that identifies the most relevant nodes and edges, (2) a spatio-temporal memory enhancement module that captures both short-term and long-term dependencies, and (3) a temporal fusion module that integrates node representations into a comprehensive global forecast. Evaluated on a real-world district heating dataset from Denmark, the SDGNN model demonstrates superior accuracy and efficiency compared to existing methods. The results indicate that the SDGNN framework effectively captures intricate spatio-temporal patterns in historical heat load data, achieving up to 5.7% improvement in RMSE, 7.4% in MAE, and 5.7% in CVRMSE over baseline models. Additionally, incorporating meteorological factors into the model further enhances its predictive performance. These findings suggest that the SDGNN framework is a robust and scalable solution for district heat load forecasting, with potential applications in other domains involving spatio-temporal graph data.

Information flow and catalytic dyad in SARS-CoV2 main protease Mpro enzyme using embedded discrete Markov chains and centrality measures

Abstract In this work, we use a network representation of the globular crystalline structure of a given protein as a graph structure with N nodes and E edges in order to analyze quantitatively the flow information and to identify key sites within the globular structure. Each node nj represents a $ C_{\alpha}^{i}$ carbon of the main backbone whereas the node’s degree ki is a measure of its physical interactions. In order to identify biologically relevant and active nodes, we compute local per residue closeness $ C_{c}(i)$, betweenness $ C_{b}(i)$ and eigenvector centralities $ C_{e}(i)$. Further analysis is done by embedding a stochastic dynamic discrete Markov chain in order to evaluate the dynamics of a set of normal random walkers (NRW’s) within the network. From this, we compute the mean first passage time matrix M and the stationary occupation probability vector ψi for each node. These two measures provide very useful information on the dynamical process embed within the $ C_{\alpha}$ network. We apply this to SARS-CoV2 Mpro main protease which is a key enzyme in the virus replication cycle. In particular, we focus our attention to the properties of the catalytic dyad integrated by His41-Cys145 in Mpro main protease since this active site has been under intense scrutiny as a pharmaceutical target. In addition, our results show the existence of additional relevant aminoacids that might play a fundamental role on signal propagation and allosteric pathways in SARS-CoV main proteases.

REFUEL: rule extraction for imbalanced neural node classification

Imbalanced graph node classification is a highly relevant and challenging problem in many real-world applications. The inherent data scarcity, a central characteristic of this task, substantially limits the performance of neural classification models driven solely by data. Given the limited instances of relevant nodes and complex graph structures, current methods fail to capture the distinct characteristics of node attributes and graph patterns within the underrepresented classes. In this article, we propose REFUEL—a novel approach for highly imbalanced node classification problems in graphs. Whereas symbolic and neural methods have complementary strengths and weaknesses when applied to such problems, REFUEL combines the power of symbolic and neural learning in a novel neural rule-extraction architecture. REFUEL captures the class semantics in the automatically extracted rule vectors. Then, REFUEL augments the graph nodes with the extracted rules vectors and adopts a Graph Attention Network-based neural node embedding, enhancing the downstream neural node representation. Our evaluation confirms the effectiveness of the proposed REFUEL approach for three real-world datasets with different minority class sizes. REFUEL achieves at least a 4% point improvement in precision on the minority classes of 1.5–2% compared to the baselines.

Improved fault detection and diagnosis using graph auto encoder and attention-based graph convolution networks

A powerful fault detection and diagnosis (FDD) system plays a pivotal role in achieving operational excellence by maximizing system performance, optimizing maintenance strategies, and ensuring the longevity and resilience of process plants. In the context of FDD for multivariate sensor data, this study presents an improved FDD approach using graph-based neural networks. This graph neural network uses an adjacency matrix developed by extracting the expert domain knowledge and topological information of the multi-sensor system. This additional graph representation of the system is incorporated along with multivariate sensor data to capture the spatial and temporal information in neural networks efficiently. In this regard, we propose and evaluate: 1) A Graph Auto Encoder (GAE) based fault detection strategy and 2) An Attention-based Spatial Temporal Graph Convolution Network (ASTGCN) based fault diagnosis methodology. By leveraging the additional knowledge in the form of graphs, the GAE captures the complex relationships and dependencies among sensors, enabling effective anomaly detection, which identifies abnormal patterns and deviations from normal behavior, thus indicating potential faults in the system. The ASTGCN incorporates attention mechanisms to selectively focus on relevant sensor nodes and capture their spatial and temporal dependencies for fault diagnosis. The effectiveness of the proposed FDD approach is demonstrated using the benchmark Tennessee Eastman Process (TEP) problem. The results show that the proposed approaches outperform traditional methods and highlight the importance of leveraging graph-based knowledge for FDD in complex systems.

Nuclear medicine procedure guideline for sentinel lymph node localization

The authors present a procedure guideline for scintigraphic detection of sentinel lymph nodes in malignant melanoma, in breast cancer, in penile and vulva tumors, in head and neck cancer, and in prostate carcinoma. Important goals of sentinel lymph node scintigraphy comprise reduction of the extent of surgery, lower postoperative morbidity and optimization of histopathological examination focussing on relevant lymph nodes. Sentinel lymph node scintigraphy itself does notdiagnose tumorous lymph node involvement and is not indicated when lymph node or distant metastases have been definitely diagnosed before sentinel lymph node scintigraphy. Procedures are compiled with the aim to reliably localise sentinel lymph nodes with a high detection rate typically in early tumour stages. New aspects in this guideline are new radiopharmaceuticals such as tilmanocept and Tc-99m-PSMA and SPECT/CT allowing an easier anatomical orientation. Initial dynamic lymphoscintigraphy in breast cancer is of little significance nowadays. Radiation exposure is low so that pregnancy is not a contraindication for sentinel lymph node scintigraphy. A one-day protocol should preferentially be used. Even with high volumes of scintigraphic sentinel lymph node procedures surgeons, theatre staff and pathologists receive a radiation exposure < 1 mSv/year so that they do not require occupational radiation surveillance. Aspects of quality control were included (scintigraphy, quality control of gamma probe, 6 h SLN course for surgeons, certified breast centers, medical surveillance center).

Recognising pedestrian behaviour using a multi-channel spatiotemporal fusion network

To protect public safety, video abnormal behaviour detection is essential. The low accuracy of identifying abnormal behaviour in pedestrians is the focus of this study. To address this, an abnormal target identification method based on multi-feature fusion of trajectory skeleton is proposed. The process begins with defining the type of abnormal behaviour in accordance with the environmental requirements. Next, pedestrian identification is carried out in the designated area, the pedestrian is tracked to determine its movement trajectory, and the image coordinates of the corresponding relevant nodes are calculated by analysing the human posture. Lastly, the track and skeleton features are integrated to classify the normal and abnormal behaviours, and the identification of abnormal target behaviours is accomplished. Test results on the behavioural analysis database indicate that this algorithm's accuracy is 87.08%, which is higher than the single-feature identification method's detection efficiency of abnormal behaviours.

Generation-based Multi-view Contrast for Self-supervised Graph Representation Learning

Graph contrastive learning has made remarkable achievements in the self-supervised representation learning of graph-structured data. By employing perturbation function (i.e., perturbation on the nodes or edges of graph), most graph contrastive learning methods construct contrastive samples on the original graph. However, the perturbation-based data augmentation methods randomly change the inherent information (e.g., attributes or structures) of the graph. Therefore, after nodes embedding on the perturbed graph, we cannot guarantee the validity of the contrastive samples as well as the learned performance of graph contrastive learning. To this end, in this article, we propose a novel generation-based multi-view contrastive learning framework (GMVC) for self-supervised graph representation learning, which generates the contrastive samples based on our generator rather than perturbation function. Specifically, after nodes embedding on the original graph we first employ random walk in the neighborhood to develop multiple relevant node sequences for each anchor node. We then utilize the transformer to generate the representations of relevant contrastive samples of anchor node based on the features and structures of the sampled node sequences. Finally, by maximizing the consistency between the anchor view and the generated views, we force the model to effectively encode graph information into nodes embeddings. We perform extensive experiments of node classification and link prediction tasks on eight benchmark datasets, which verify the effectiveness of our generation-based multi-view graph contrastive learning method.

GC-CDSS: Personalized gastric cancer treatment recommendations system based on knowledge graph

BackgroundGastric cancer (GC) is one of the most common malignant tumors in the world, posing a serious threat to human health. Currently, gastric cancer treatment strategies emphasize a multidisciplinary team (MDT) consultation approach. However, there are numerous treatment guidelines and insights from clinical trials. The application of AI-based Clinical Decision Support System (CDSS) in tumor diagnosis and screening is increasing rapidly. ObjectiveThe purpose of this study is to (1) summarize the treatment decision process for GC according to the treatment guidelines in China, and then create a knowledge graph (KG) for GC, (2) based on aforementioned KG, built a CDSS and conducted an initial feasibility evaluation for the current system. MethodsFirstly, we summarized the decision-making process for treatment of GC. Then, we extracted relevant decision nodes and relationships and utilized Neo4j to create the KG. After obtaining the initial node features for building the graph embedding model, graph embedding algorithm, such as Node2Vec and GraphSAGE, were used to construct the GC-CDSS. At last, a retrospective cohort study was used to compare the consistency between GC-CDSS and MDT in treatment decision making. ResultsIn current study, we introduce a GC-CDSS, which is constructed based on Chinese GC treatment guidelines knowledge graph (KG). In the KG, we define four types of nodes and four types of relationships, and it comprise a total of 207 nodes and 300 relationships. Regarding GC-CDSS, the system is capable of providing dynamic and personalized diagnostic and treatment recommendations based on the patient's condition. Furthermore, a retrospective cohort study is conducted to compare GC-CDSS recommendations with those of the MDT group, the overall consistency rate of treatment recommendations between the auxiliary decision system and MDT team is 92.96%. ConclusionsWe construct a GC treatment support system, GC-CDSS, based on KG. The GC-CDSS may help oncologists make treatment decisions more efficient and promote standardization in primary healthcare settings.

Saliency-aware regularized graph neural network

The crux of graph classification lies in the effective representation learning for the entire graph. Typical graph neural networks focus on modeling the local dependencies when aggregating features of neighboring nodes, and obtain the representation for the entire graph by aggregating node features. Such methods have two potential limitations: 1) the global node saliency w.r.t. graph classification is not explicitly modeled, which is crucial since different nodes may have different semantic relevance to graph classification; 2) the graph representation directly aggregated from node features may have limited effectiveness to reflect graph-level information. In this work, we propose the Saliency-Aware Regularized Graph Neural Network (SAR-GNN) for graph classification, which consists of two core modules: 1) a traditional graph neural network serving as the backbone for learning node features and 2) the Graph Neural Memory designed to distill a compact graph representation from node features of the backbone. We first estimate the global node saliency by measuring the semantic similarity between the compact graph representation and node features. Then the learned saliency distribution is leveraged to regularize the neighborhood aggregation of the backbone, which facilitates the message passing of features for salient nodes and suppresses the less relevant nodes. Thus, our model can learn more effective graph representation. We demonstrate the merits of SAR-GNN by extensive experiments on seven datasets across various types of graph data.

Empowering 6G maritime communications with distributed intelligence and over-the-air model sharing

Introduction: Shipping and maritime transportation have gradually gained a key role in worldwide economical strategies and modern business models. The realization of Smart Shipping (SMS) powered by advanced 6G communication networks, as well as innovative Machine Learning (ML) solutions, has recently become the focal point in the maritime sector. However, conventional centralized learning schemes are unsuitable in the maritime domain, due to considerable data communication overhead, stringent energy constraints, increased transmission failures in the harsh propagation environment, as well as data privacy concerns.Methods: To overcome these challenges, we propose the joint adoption of Federated Learning (FL) principles and the utilization of the Over-the-Air computation (AirComp) wireless transmission framework. Thus, this paper initially describes the mathematical considerations of a 6G maritime communication system, focusing on the heterogeneity of the relevant nodes and the channel models, including an Unmanned Aerial Vehicle (UAV)-aided relaying model that is usually required in maritime communications. The communication network, enhanced with the AirComp technique for efficiency purposes, forms the technical basis for the collaborative learning across multiple Internet of Maritime Things (IoMT) nodes in FL tasks. The workflow of the FL/AirComp scheme is illustrated and proposed as a communication-efficient and privacy-aware SMS framework, considering spectrum and energy efficiency aspects under a sum transmitting power constraint.Results: Then, the performance of the proposed methodology is assessed in an important ML task, related to intelligent maritime transportation systems, namely, the prediction of the Cargo Ship Propulsion Power using real data originating from six cargo ships and utilizing long-short-term-memory (LSTM) neural networks. Upon extensive experimentation, FL showed higher prediction accuracy relative to the typical Ensemble Learning technique by a factor of 3.04. The AirComp system performance was evaluated under varying noise conditions and number of IoMT nodes, using simulation data for the channel state information by regulating the power of the transmitting IoMT entities and the scaling factor at the shore base station.Discussion: The results clearly indicate the efficiency of the proposed FL/AirComp scheme in achieving low computation error, collaborative learning, spectrum efficiency and privacy protection in wireless maritime communications, while providing adequate accuracy levels with respect to the optimization objective.

A Network Approach of Gambling Disorder Profile with and Without Related Illegal Acts

AbstractGambling disorder (GD) is a complex mental health condition that can cause many severe psychological, physical, and social impairment. Illegal acts have been recognized in quite a few cases because of the debts related with the gambling activity. This study used network methodology to visualize the relationships among patients seeking treatment for gambling related problems, separately for the patients with and without illegal behaviors. The aim is to identify the diverse and differentiate mechanisms, as well as the central nodes, that occur within GD patients depending on the presence/absence of illegal acts. The sample included N = 401 patients (age range 18 to 80 years). Network analysis was performed considering the nodes that measure gambling features (the core symptoms based on the DSM-5 taxonomy, global symptom severity, and forms of gambling), psychopathology distress, substance use (tobacco, alcohol, and illegal drugs), and personality traits. Two separate networks were adjusted for patients with illegal acts (n = 105) and without these behaviors (n = 296). The most relevant nodes among patients with GD plus illegal acts were self-transcendence and the GD DSM-5 symptom “A7-lies to conceal the extent of gambling” (these variables were also identified as the bridge nodes, those with the highest linkage capacity). Among the patients with GD without illegal acts, the node with the greatest authority was the GD DSM-5 symptom “A5-often gambles when feeling distressed” (this was also the variable with the highest linkage capacity). The study provides empirical evidence of the most relevant features and the linkage capacity among patients seeking treatment for problematic gambling, which can support the development of precise plans for treatment and prevention of the risk of GDRIA.

Towards self-explainable graph convolutional neural network with frequency adaptive inception

Graph convolutional neural networks (GCNs) have demonstrated powerful representing ability of irregular data, e.g., skeletal data and graph-structured data, providing the effective mechanism to fuse the neighbor nodes. However, inheriting from the deep learning, GCN also lacks interpretability, which hinders its application to scenarios that have high demand for transparency. Although, there have been many efforts on the interpretability of deep learning, they mainly concentrate on i.i.d data that is hard to be deployed to GCNs, which involve not only the node feature, but also the graph structure. There are few works that attempt to explain it with post-hoc manner, which can be biased, resulting in mis-representation of the true explanation. Therefore, in this paper, we propose a framework, namely ExpFiGCN, that reveals explainability of the GCNs from the perspective of graph structure and mathematical analysis. Specifically, ExpFiGCN can find the most intrinsically relevant node to the central node and obtain the informative and discriminative signals while performing denoising. For the graph structure, we find K-nearest nodes; for the mathematical analysis, every channel of a node and its neighborhoods contribute dynamically to the final channel signal, which can capture the inherent difference of different channels and neighbor nodes. Meanwhile, it can enhance the representation ability of nodes and ameliorate the over-smoothing problem. On the other hand, our model can dynamically adjust the importance of neighborhoods to the central vertex. We empirically validate the effectiveness of the proposed framework ExpFiGCN on various benchmark datasets. Experimental results show that our method achieves substantial improvements and outperforms the state-of-the-art performance strikingly.

Graph neural network method for the intelligent selection of river system

The spatial features and generalisation rules for river network generalisation are difficult to directly quantify using indicators. To consider dimensional information hidden in river networks and improve river network selection accuracy, this study introduces a graph convolutional neural network-based method. First, we modelled the river network as a graph structure, where the nodes represent each river segment and edges represent the connections between river segments. The semantic, geometric, and morphological features of individual river segments and topological and constraint features between river segments were then calculated to characterise the relevant nodes. Second, under supervised classification, the input node attributes and labels were sampled and aggregated to obtain richer and more abstract high-level features. The graph convolutional neural network model then selected or deleted river segments. Finally, the selected individual river segments were connected to obtain the complete integrated river network. A 1:10,000 scale map of the Min River system in the Yangtze River Basin was tested, with a 1:50,000 scale map used as the control, and it yielded a correct classification rate >95%. Moreover, the correct classification rate was 7.35%–5.31% and 7.7%–3.3% higher than that of other graph neural network methods and traditional machine learning methods, respectively.

An integrated approach for depression diagnosis using 3S feature embeddings and G-BLS with T-pHBGO optimizer

Mental health is an integral part of human welfare in the present era. The present research aims to bring attention to mental health issues with the help of social network, which turns into an influential forum for depressed individuals to express their sentiments and emotions. In this work, a contemporary depression detection framework that incorporates a prominent feature extraction approach and an optimized Gated Broad Learning System (G-BLS) is proposed. The feature extraction method captures Syntactic, Semantic, and Sentiment information of words and forms 3S feature embeddings. The extracted 3S feature vectors are then fed into the G-BLS model for depressive tweet classification. The G-BLS network often transfers redundant hidden nodes to the output layer because of its structural properties. To address this problem, a new binary optimization algorithm abbreviated as T-pHBGO is introduced by hybridizing the Honey Badger (HB) algorithm and Gannet Optimizer (GO) along with a T-shaped binary transfer function and population reduction strategy. The T-pHBGO algorithm selects the relevant and ideal neuron nodes to intensify the classification performance. The extensive experiments are conducted using two distinct Twitter corpora from the Kaggle repository. Accuracy and Root Mean Square Error (RMSE) measures are used to evaluate the system in contrast to other existing depression detection models. The comparative outcomes revealed that the composition of 3S word embedding, G-BLS model, and T-pHBGO algorithm surpasses the competing methods with an accuracy of 99.2% and 96.3%, RMSE rate of 0.0592 and 0.0417 for Sentiment 140 and Twitter depression corpus. The Analysis of Variance F-test confirmed the statistical significance of the results. The cross-validation results demonstrate that the proposed method is capable of maintaining its performance despite the use of different test data, as evidenced by the average classification accuracy.

Dynamic Probabilistic Risk Assessment of Overpressure Burst of Pipeline with Pressure-Protection Systems in Case of Blockage

Although overpressure burst of pipeline with pressure-protection systems occurs infrequently in case of blockage, it endangers personnel safety and causes serious economic loss when it happens. Therefore, assessing the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage is necessary to prevent accidents. This study proposes a method based on a dynamic Bayesian network (DBN) for assessing the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage. First, a fault tree (FT) model of pipeline overpressure burst is constructed. The constructed FT model is then mapped into the DBN model to solve the uncertainty of the model. Second, the leaky Noisy-OR gate is used to define the uncertain logical relationships of relevant nodes in the DBN model. The effect of common cause failures on a pipeline’s redundant pressure-protection system is analyzed using the multiple error shock model. Finally, a natural gas export pipeline is presented as an example to demonstrate the proposed method. Results show that the method can effectively assess the dynamic probabilistic risk of overpressure burst of pipeline with pressure-protection systems in case of blockage.

ENHANCING MELANOMA CLASSIFICATION WITH GRAPH ATTENTION LAYERS AND GROUP METHOD OF DATA HANDLING - BASED FEATURE EXTRACTION

Melanoma, a deadly form of skin cancer, demands accurate and early diagnosis for effective treatment. In this study, we propose a novel approach to improve melanoma classification by integrating Graph Attention Layers (GALs) into the Group Method of Data Handling (GMDH) framework. Our method leverages the power of GMDH to automatically generate and select informative features from complex melanoma-related data. Simultaneously, GALs are employed to capture intricate relationships and dependencies within the data, often overlooked by traditional classification models. We construct a graph representation where nodes represent data elements (patients or genetic markers) and edges signify relationships between them. GALs are applied to the graph, allowing the model to attend to relevant nodes and connections, enhancing its ability to discern subtle patterns indicative of melanoma. We then train a classification model on this enriched feature set, aiming for superior accuracy in melanoma diagnosis. Experimental results on a diverse melanoma dataset demonstrate the effectiveness of our approach. The model consistently outperforms traditional methods in terms of accuracy, precision, and recall. This study highlights the potential of combining GMDH-based feature extraction with GALs in melanoma classification. This approach not only advances diagnostic accuracy but also provides valuable insights into the underlying factors driving melanoma risk. As early detection remains the key to melanoma treatment success, our proposed method holds promise for improving patient outcomes.

Efficient search for relevance explanations using MAP-independence in Bayesian networks

▪-independence is a novel concept concerned with explaining the (ir)relevance of intermediate nodes for maximum a posteriori (▪) computations in Bayesian networks. Building upon properties of ▪-independence, we introduce and experiment with methods for finding sets of relevant nodes using both an exhaustive and a heuristic approach. Our experiments show that these properties significantly speed up run time for both approaches. In addition, we link ▪-independence to defeasible reasoning, a type of reasoning that analyses how new evidence may invalidate an already established conclusion. Ways to present users with an explanation using ▪-independence are also suggested.