Neighbors In Network Research Articles

The evolution of cooperation and global synchronization in the evolutionary Kuramoto dilemma combined with the prisoner's dilemma

In this work, we have studied the evolution of the cooperation and the global synchronization on the Kuramoto model upon random networks with two games, in which one is the evolutionary Kuramoto dilemma and the other is the weak prisoner's dilemma. Agents can gain their payoffs from these two games synchronously. In the evolutionary Kuramoto game, cooperators prefer the formation of local synchronization but pay less cooperative cost. Cooperators in weak prisoner's dilemma can form cooperator clusters because of the network reciprocity. Combining these two games shows a mutual promotion effect upon the order of the global synchronization but a competition effect on the cooperation level. The global synchronization begins to rise only when most agents choose cooperation. It is manifested by the fact that the optimal coupling strength for the highest cooperation level is smaller than that for the highest synchronization level. The high cooperation cost makes defection the most advantageous strategy when the coupling strength reaches rather high, which diminishes the effect of cooperator clusters, and ultimately results in the absence of both cooperators and global synchronization. When agents have more neighbors in the random network, it is observed that cooperation is consistently hindered unless the coupling strength is relatively small. Additionally, there is always an optimal average degree that allows for global synchronization.

Network formation and efficiency in linear-quadratic games: An experimental study

Abstract We experimentally study effort provision and network formation in the linear-quadratic game characterised by positive externality and complementarity of effort choices among network neighbours. We compare experimental outcomes to the equilibrium and efficient allocations and study the impact of group size and linking costs. We find that individuals overprovide effort relative to the equilibrium level on the network they form. However, their payoffs are lower than the equilibrium payoffs because they create fewer links than is optimal, which limits the beneficial spillover effects of effort provision. Reducing the linking costs does not significantly increase the connectedness of the network and the welfare loss is higher in larger groups. Individuals connect to the highest-effort providers in the group and ignore links to relative low-effort providers, even if those links would be beneficial to form. This effect explains the lack of links in the network.

Advancing architectural frameworks for vibration signature classification in rotating machinery

Advancements in data-driven predictive maintenance have significantly improved digital twin applications for rotating machinery, offering robust solutions for smart manufacturing challenges. These improvements are crucial since equipment failures can cause extensive and costly disruptions to both maintenance schedules and operations. As precision and reliability are critical in production processes, undetected fluctuations in operating frequencies can swiftly escalate to complete part failure, leading to prolonged repairs and productivity loss. This study explores an integrated dataflow pipeline, specifically through Siemens’ MindSphere, to enable continuous predictive maintenance and enhance data acquisition and management. Particularly, conditions such as normal operation, mass balance, rotating imbalance, and mechanical looseness are classified using support vector machine (SVM), neural network (NN), and K-Nearest Neighbor (KNN) methods for the purpose of comparing results. Our results highlight the efficacy of ensemble techniques in collecting and diagnosing vibration signatures, thereby enabling proactive maintenance. To classify various failure signatures, we have proposed a framework to interpret time-series and frequency-dependent data for determining failure types. This research exemplifies how merging data-driven methods with digital twin can improve the accuracy and reliability of condition monitoring. Additionally, we introduce a cloud-based architecture for the diagnosis of rotating machinery, utilizing Application Programming Interface (API) configurations, and develop a real-time dashboard for streaming and visualizing classified data, fostering immediate and informed decision-making.

Strategic Customer Segmentation: Harnessing Machine Learning For Retaining Satisfied Customers

This research paper explores the burgeoning field of machine learning and its application in strategic customer segmentation within the aviation industry. Leveraging the Airline Passenger dataset, this study assesses the potential of various machine learning classifiers to enhance customer retention by effectively segmenting satisfied customers. Our methodology involves a comparative analysis of five machine learning classifiers: Random Forest, K-Nearest Neighbors (KNN), Decision Tree, Naive Bayes, and Artificial Neural Network (ANN). Each classifier is rigorously tested and evaluated based on key performance metrics including accuracy, precision, recall, and F1-score. The results indicate a diverse range of classifier effectiveness. Notably, the Random Forest classifier outperforms others with outstanding metrics: accuracy, precision, recall, and F1-scoreof 0.96. Decision Tree follows closely, also achieving high performance with a score of 0.95 across all metrics. Naive Bayes and ANN demonstrate respectable performance, with accuracyscores of 0.86 and 0.90 respectively. In contrast, KNN presents lower but consistent performance, with all metrics at 0.75. These quantitative findings highlight the nuanced performancedifferences among classifiers, emphasizing the critical role of algorithm selection in achieving precise customer segmentation. This study provides significant insights into the application of machine learning for strategic customer retention in the aviation sector, presenting practical implications for airlines aiming to optimize their segmentation strategies and retain satisfied customers. By showcasing the varying performances of different classifiers, this research contributes to the broader discourse on integrating machine learning into customer-centric strategies, ultimately aiding airlines to engage and retain their customer base more effectively.

On the friendship paradox and inversity: A network property with applications to privacy-sensitive network interventions

We provide the mathematical and empirical foundations of the friendship paradox in networks, often stated as "Your friends have more friends than you." We prove a set of network properties on friends of friends and characterize the concepts of ego-based and alter-based means. We propose a network property called inversity that quantifies the imbalance in degrees across edges and prove that the sign of inversity determines the ordering between ego-based or alter-based means for any network, with implications for interventions. Network intervention problems like immunization benefit from using highly connected nodes. We characterize two intervention strategies based on the friendship paradox to obtain such nodes, with the alter-based and ego-based strategy. Both strategies provide provably guaranteed improvements for any network structure with variation in node degrees. We demonstrate that the proposed strategies obtain several-fold improvement (100-fold in some networks) in node degree relative to a random benchmark, for both generated and real networks. We evaluate how inversity informs which strategy works better based on network topology and show how network aggregation can alter inversity. We illustrate how the strategies can be used to control contagion of an epidemic spreading across a set of village networks, finding that these strategies require far fewer nodes to be immunized (less than 50%, relative to random). The interventions do not require knowledge of network structure, are privacy-sensitive, are flexible for time-sensitive action, and only require selected nodes to nominate network neighbors.

Enhancing local road pavement condition prediction using Bayesian-optimized ensemble machine learning and adaptive synthetic sampling technique

ABSTRACT Pavement condition prediction helps road agencies to schedule maintenance, rehabilitation, and reconstruction, and to allocate limited funds and resources to such activities. Compared to state highways, pavement performance prediction for local roads has received relatively little attention in the literature due to perceptions of low importance, low levels of investment in data collection, poor data quality, and high variation within the data elements. Additionally, local road pavement condition data may suffer from dataset imbalance, often leading to unreliable condition predictions. Hence, this paper introduces a methodology to predict local pavement condition using various single estimator and ensemble machine learning (ML) models along with the adaptive synthetic sampling method. The study develops nine (9) Bayesian-optimised ML models: category boosting (CatBoost), adaptive boosting, decision tree, extra trees, gradient boosting, light gradient-boosting machine, k-nearest neighbour, random forest, and artificial neural network. The ensemble ML and CatBoost were found to exhibit the best model performance, with an average testing accuracy of 0.82, specificity of 0.81, sensitivity of 0.63, and F-measure of 0.61. These results underscore the efficacy of ensemble ML models in pavement condition prediction. The proposed approach can be beneficial to local road agencies in their long-term planning, scheduling, and budgeting.

T-distributed Stochastic Neighbor Network for unsupervised representation learning

Unsupervised representation learning (URL) is still lack of a reasonable operator (e.g. convolution kernel) for exploring meaningful structural information from generic data including vector, image and tabular data. In this paper, we propose a simple end-to-end T-distributed Stochastic Neighbor Network (TsNet) for URL with clustering downstream task. Concretely, our TsNet model has three major components: (1) an adaptive connectivity distribution learning module is presented to construct a pairwise graph for preserving the local structure of generic data; (2) a T-distributed stochastic neighbor embedding based loss function is designed to learn a transformation between embeddings and original data, which improves the discrimination of representations; (3) a nonlinear parametric mapping is learned via our TsNet on an unsupervised generalized manner, which can address the “out-of-sample” issue. By combining these components, our method is able to considerably outperform previous related unsupervised learning approaches on visualization and clustering of generic data. A simple deep neural network equipped on our model respectively achieves 74.90%, 76.56% ACC and NMI, which is 8% relative improvement over previous state-of-the-art on real single-cell RNA-sequencing (scRNA-seq) datasets clustering.

Development of neighborhood networks in suburban areas for persuading women to adhere to breast cancer screening programs

Introduction: The purpose of this study was to determine the effectiveness of the development of neighborhood networks in suburban areas for increasing adherence to breast cancer screening. Methods: This community-based intervention study was conducted among Iranian urban women aged 30 to 65 in 2021. The samples were chosen using a multi-stage random sampling process. Women were randomly assigned to intervention group (n=271) and control group (n=155). Following a three-month intervention period, participants were asked about their breast cancer screening behaviors to gauge the program’s success. Cognitive factors including perceived social support, perceived benefits and perceived barriers toward breast cancer screening, subjective norms, behavioral intention, and breast cancer screening behaviors were collected. The data were analyzed with SPSS statistical software version 21. Results: After the intervention, the results showed an improvement in the intention of referring to mammography centers, perceived benefits of early diagnosis of breast cancer, receiving social support, breast cancer screening behaviors, and a decrease in barriers to early diagnosis of breast cancer compared to the control group (P<0.05). Conclusion: Harnessing the power of social networks alongside educational initiatives and strong community support can provide a safe, affordable, and effective means of boosting participation in Breast Cancer Screening (BCS).

How can a community be successfully empowered to deal with death, dying, and bereavement?-formative evaluation of the Caring Community Cologne using focus groups.

Compassionate communities aim to empower people to deal with death, dying, and bereavement. They also intend to facilitate access to care and support at the end of life. However, there is a need for systematic knowledge on how to achieve the desired outcomes for citizens and for insights regarding the development, implementation, and evaluation. The aim of this study was to assess the views of members of a German Compassionate City, the "Caring Community Cologne" (CCC), and to report on its practical implementation. The CCC consists of a citywide Round Table, a Steering Group, a Coordination Office and four Working Groups in areas where activities are already in place. We conducted two qualitative focus groups with nine members of three Working Groups. The transcripts were analysed with qualitative content analysis, using MAXQDA version 2022, and results were transferred into the logic model "Throughput Model". At the time of evaluation, participants felt that the structures of the CCC were adequate, but criticised the cooperation and transparency between them. A key aspect of this was the requirement for a coordinating body. They stressed the support of federal institutions as a key factor, while at the same time describing insufficient citizen involvement. The transfer of the results into the Throughput Model highlighted four areas that the CCC should address: (I) neighbourhood networks need to be established to strengthen civic support; (II) people need to be made aware of the issues by making them accessible in their everyday lives; (III) the many existing support initiatives need to be better linked and made more accessible; (IV) adequate healthcare service structures have to be guaranteed. The top-down approach described, supported by the city's engagement and involving existing initiatives can facilitate the development of a bottom-up civic engagement model in a large city. However, active citizen involvement appeared to be a challenge. The Throughput Model was a suitable basis for mapping work processes and developing evaluation plans.

Adaptive networks driven by partner choice can facilitate coordination among humans in the graph coloring game: Evidence from a network experiment

Background Many instances of coordination problems in social networks can be characterized by the well-known graph coloring game—a task in which agents are incentivized to choose an attribute, represented by a color, to be different from linked neighbors. Past work on this topic, however, is limited to static networks of specific topologies of interest. Purpose This study aims to explore whether dynamic networks, where players can choose their partners, perform better in the graph coloring game. Research Design We replicate the conventional graph coloring game experiment and introduce a new version where players can change their network neighbors. Study Sample and Data Collection We recruited over 300 human participants for the experiment, with each session involving an average of 20 participants per network. Results We found that dynamic networks that allow agents to switch network neighbors works more effectively and efficiently in solving coloring conflicts than static networks in which people are allowed to change color alone. The networks in the experiment updated by people’s selection of neighbors tend to evolve to a state similar in structure to the small-world network characterized by short path distance and high triadic clustering. Conclusion We argue that a self-organized approach demonstrated by our experiment can inspire search of behavioral strategies and network topologies for solving coordination problems in social networks.

Machine learning approaches for asthma disease prediction among adults in Sri Lanka.

Objectives: Addressing the challenge of cost-effective asthma diagnosis amidst diverse symptom patterns among patients, this study aims to develop a machine learning-based asthma prediction tool for self-detection of asthma. Methods: Data from 6,665 participants in the Sri Lanka Health and Ageing Study (2018-2019) are used for this research. Thirteen machine learning algorithms, including Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, Naïve Bayes, K-Nearest Neighbors, Gradient Boost, XGBoost, AdaBoost, CatBoost, LightGBM, Multi-Layer Perceptron, and Probabilistic Neural Network, are employed. Results: A hybrid version of Logistic Regression and LightGBM outperformed other models, achieving an AUC of 0.9062 and 79.85% sensitivity. Key predictive features for asthma include wheezing, breathlessness with wheezing, shortness of breath attacks, coughing attacks, chest tightness, nasal allergies, physical activity, passive smoking, ethnicity, and residential sector. Conclusion: Combining Logistic Regression and LightGBM models can effectively predict adult asthma based on self-reported symptoms and demographic and behavioural characteristics. The proposed expert system assists clinicians and patients in diagnosing potential asthma cases.

SFINN: inferring gene regulatory network from single-cell and spatial transcriptomic data with shared factor neighborhood and integrated neural network.

The rise of single-cell RNA sequencing (scRNA-seq) technology presents new opportunities for constructing detailed cell type-specific gene regulatory networks (GRNs) to study cell heterogeneity. However, challenges caused by noises, technical errors, and dropout phenomena in scRNA-seq data pose significant obstacles to GRN inference, making the design of accurate GRN inference algorithms still essential. The recent growth of both single-cell and spatial transcriptomic sequencing data enables the development of supervised deep learning methods to infer GRNs on these diverse single-cell datasets. In this study, we introduce a novel deep learning framework based on shared factor neighborhood and integrated neural network (SFINN) for inferring potential interactions and causalities between transcription factors and target genes from single-cell and spatial transcriptomic data. SFINN utilizes shared factor neighborhood to construct cellular neighborhood network based on gene expression data and additionally integrates cellular network generated from spatial location information. Subsequently, the cell adjacency matrix and gene pair expression are fed into an integrated neural network framework consisting of a graph convolutional neural network and a fully-connected neural network to determine whether the genes interact. Performance evaluation in the tasks of gene interaction and causality prediction against the existing GRN reconstruction algorithms demonstrates the usability and competitiveness of SFINN across different kinds of data. SFINN can be applied to infer GRNs from conventional single-cell sequencing data and spatial transcriptomic data. SFINN can be accessed at GitHub: https://github.com/JGuan-lab/SFINN.

Optimasi Algoritma Machine Learning Menggunakan Teknik Bagging Pada Klasifikasi Diagnosis Kanker Payudara

Classification algorithms have a very important role in Machine Learning, but not all algorithms have the same performance in every case. Algorithm performance can be affected by the type of data used, differences in problem characteristics, and the parameters used. Additionally, ensemble learning techniques such as Bagging can affect algorithm performance. Therefore, the problem arises of how to choose the most suitable algorithm for a particular classification task and how to optimize the performance of the algorithm. This research aims to carry out a comparative analysis and optimization of classification algorithms in Machine Learning. Classification algorithms that will be evaluated include Support Vector Machine (SVM), Neural Network, Logistic Regression, Decision Tree, and K-Nearest Neighbors (K-NN). Evaluation of the performance of these algorithms will be carried out using the confusion matrix, Receiver Operating Characteristic (ROC) Curve, and Area Under Curva (AUC). The result of this research is a comparative analysis of the optimization of classification algorithms using the bagging technique. After carrying out the evaluation process using the confusion matrix and ROC curve, it was found that the algorithm optimization using the bagging technique only had an effect on the Decision Tree (DT) and K-Nearest Neighbors (KNN) algorithms. . The accuracy of the DT algorithm increased by 0.6% while the accuracy of KNN increased by 1.3%. The AUC value for the DT algorithm increased by 1.4% and the KNN algorithm increased by 0.3%.

The health implications of neighborhood networks based on daily mobility in US cities

A large body of research has been dedicated to understanding the neighborhood conditions that impact health, which outcomes are affected, and how these associations vary by demographic and socioeconomic neighborhood and individual characteristics. This literature has focused mostly on the neighborhoods in which individuals reside, thus failing to recognize that residents across race/ethnicity and class spend a non-trivial amount of their time in neighborhoods far from their residential settings. To address this gap, we use mobile phone data from the company SafeGraph to compare racial inequality in neighborhood socioeconomic advantage exposure across three scales: the neighborhoods that residents live in, their adjacent neighborhoods, and the neighborhoods that they regularly visit. We found that the socioeconomic advantage levels in neighborhood networks differ from the levels at the residential and adjacent scales across all ethnoracial neighborhoods. Furthermore, socioeconomic advantage at the network level is associated with diabetes and hypertension prevalence above and beyond its impact at the residential and adjacent levels. We also find ethnoracial differences in these associations, with greater beneficial consequences of network socioeconomic advantage exposure on hypertension and diabetes for white neighborhoods. Future social determinants of health research needs to reconceptualize exposure to include the larger neighborhood network that a community is embedded in based on where their residents travel to and from.

Predictive Study on the Occurrence of Wheat Blossom Midges Based on Gene Expression Programming with Support Vector Machines.

This study addresses the challenges in plant pest and disease prediction within the context of smart agriculture, highlighting the need for efficient data processing techniques. In response to the limitations of existing models, which are characterized by slow training speeds and a low prediction accuracy, we introduce an innovative prediction method that integrates gene expression programming (GEP) with support vector machines (SVM). Our approach, the gene expression programming-support vector machine (GEP-SVM) model, begins with encoding and fitness function determination, progressing through cycles of selection, crossover, mutation, and the application of a convergence criterion. This method uniquely employs individual gene values as parameters for SVM, optimizing them through a grid search technique to refine genetic parameters. We tested this model using historical data on wheat blossom midges in Shaanxi Province, spanning from 1933 to 2010, and compared its performance against traditional methods, such as GEP, SVM, naive Bayes, K-nearest neighbor, and BP neural networks. Our findings reveal that the GEP-SVM model achieves a leading back-generation accuracy rate of 90.83%, demonstrating superior generalization and fitting capabilities. These results not only enhance the computational efficiency of pest and disease prediction in agriculture but also provide a scientific foundation for future predictive endeavors, contributing significantly to the optimization of agricultural production strategies.

Predicting Success in Descemet Membrane Endothelial Keratoplasty Using Machine Learning.

This study aimed to predict early graft failure (GF) in patients who underwent Descemet membrane endothelial keratoplasty based on donor characteristics. Several machine learning methods were trained to predict GF automatically. To predict GF, the following variables were obtained: donor age, sex, systemic diseases, medications, duration of stay in the intensive care unit, death-to-preservation time (DPT), endothelial cell density of the cornea, tightness of Descemet membrane roll during surgery, anterior chamber tamponade, tamponade used for rebubbling, and preoperative best corrected visual acuity. Five classification methods were experimented with the study data set: random forest, support vector machine, k-nearest neighbor, RUSBoosted tree, and neural networks. In holdout validation, 75% of the data were used in training and the remaining 25% used in testing. The predictive accuracy, sensitivity, specificity, f-score, and area under the receiver operating characteristic curve of the methods were evaluated. The highest classification accuracy achieved during the experiments was 96%. The precision, recall, and f1-score values were 0.95, 0.81, and 0.90, respectively. Feature importance was also computed using analysis of variance. The model revealed that GF risk was related to DPT and the intensive care unit duration ( P < 0.05). No significant relationship was found between donor age, endothelial cell density, systemic diseases and medications, graft roll, tamponades, and GF risk. This study shows a strong relationship between increased intensive care duration, DPT, and GF. Experimental results demonstrate that machine learning methods may effectively predict GF automatically.

Machine learning-based thermal performance study of microchannel heat sink under non-uniform heat load conditions

The parallel microchannel heat sink stands as a pivotal solution in managing high heat flux electronics due to its efficient heat transfer characteristics and ease of manufacturing. While numerous studies have explored the thermal performance and flow characteristics of microchannel heat sinks, most have focused on uniform heat loads or relied heavily on numerical methods. This study presents an experimental system tailored to generate data for analyzing the thermal performance of microchannel heat sinks under various conditions. Leveraging this dataset, four distinct machine learning models Artificial Neural Network (ANN), XGBoost, LightGBM, and K-nearest neighbor (KNN) were trained using 22 input features, totalling 560 data points categorised into geometry parameters, heating patterns, and boundary conditions details. The models were tasked with predicting six response variables: the average base temperature of the heat sink, temperature change (ΔT), hotspot temperature, heat transfer coefficient (h), Nusselt number (Nu), and thermal resistance (Rth). Among the four machine learning models, XGBoost exhibited a good predictive accuracy of an average R2 value of 0.98 and MAE values of 2.1 across all responses. Furthermore, the study delved into the impact of varying input features on prediction accuracy, revealing a consistent enhancement in accuracy with the inclusion of more features across all models.

Does public rental housing foster social ties? A study of the everyday social lives of rural migrants in Chongqing, China

ABSTRACT Public housing is a crucial means of providing sufficient housing solutions worldwide. However, many experts criticize the harsh architecture, the absence of public space, and lacking social infrastructure in these densely populated public housing projects in the Global North. In China, previous research has mainly focused on the effect of public rental housing (PRH) on real estate values, employment trajectories of rural migrants, and their life challenges. Few studies have explored the dynamic relationship between rural migrants, their daily social life, and how it interacts with the environment. By conducting multiple site observations and 120 semi-structured interviews with rural migrants living in three different PRH sites in Chongqing, this study addresses these gaps and uncovers the complexities of their daily social lives. The findings show that rural migrant residents have established neighborhood networks, provided mutual support, and developed a sense of place attachment to their new homes through various strategies in their daily social lives. The research enriches the theoretical debate on migration theories and deepens our understanding of rural migrants’ social experiences in urban settings. It also sheds light on China’s sustainable urban development and the overall well-being of migrating populations.

Detection of Parkinson disease using multiclass machine learning approach

Parkinson’s Disease (PD) is a prevalent neurological condition characterized by motor and cognitive impairments, typically manifesting around the age of 50 and presenting symptoms such as gait difficulties and speech impairments. Although a cure remains elusive, symptom management through medication is possible. Timely detection is pivotal for effective disease management. In this study, we leverage Machine Learning (ML) and Deep Learning (DL) techniques, specifically K-Nearest Neighbor (KNN) and Feed-forward Neural Network (FNN) models, to differentiate between individuals with PD and healthy individuals based on voice signal characteristics. Our dataset, sourced from the University of California at Irvine (UCI), comprises 195 voice recordings collected from 31 patients. To optimize model performance, we employ various strategies including Synthetic Minority Over-sampling Technique (SMOTE) for addressing class imbalance, Feature Selection to identify the most relevant features, and hyperparameter tuning using RandomizedSearchCV. Our experimentation reveals that the FNN and KSVM models, trained on an 80–20 split of the dataset for training and testing respectively, yield the most promising results. The FNN model achieves an impressive overall accuracy of 99.11%, with 98.78% recall, 99.96% precision, and a 99.23% f1-score. Similarly, the KSVM model demonstrates strong performance with an overall accuracy of 95.89%, recall of 96.88%, precision of 98.71%, and an f1-score of 97.62%. Overall, our study showcases the efficacy of ML and DL techniques in accurately identifying PD from voice signals, underscoring the potential for these approaches to contribute significantly to early diagnosis and intervention strategies for Parkinson’s Disease.

Network-based time series modeling for COVID-19 incidence in the Republic of Ireland

Network-based time series models have experienced a surge in popularity over the past years due to their ability to model temporal and spatial dependencies, arising from the spread of infectious disease. The generalised network autoregressive (GNAR) model conceptualises time series on the vertices of a network; it has an autoregressive component for temporal dependence and a spatial autoregressive component for dependence between neighbouring vertices in the network. Consequently, the choice of underlying network is essential. This paper assesses the performance of GNAR models on different networks in predicting COVID-19 cases for the 26 counties in the Republic of Ireland, over two distinct pandemic phases (restricted and unrestricted), characterised by inter-county movement restrictions. Ten static networks are constructed, in which vertices represent counties, and edges are built upon neighbourhood relations, such as railway lines. We find that a GNAR model based on the fairly sparse Economic hub network explains the data best for the restricted pandemic phase while the fairly dense 21-nearest neighbour network performs best for the unrestricted phase. Across phases, GNAR models have higher predictive accuracy than standard ARIMA models which ignore the network structure. For county-specific predictions, in pandemic phases with more lenient or no COVID-19 regulation, the network effect is not quite as pronounced. The results indicate some robustness to the precise network architecture as long as the densities of the networks are similar. An analysis of the residuals justifies the model assumptions for the restricted phase but raises questions regarding their validity for the unrestricted phase. While generally performing better than ARIMA models which ignore network effects, there is scope for further development of the GNAR model to better model complex infectious diseases, including COVID-19.