Network Proximity Research Articles

P-RAN: A Distributed Solution for Cellular Systems in High Frequency Bands

To evolve to the next generation, cellular systems are expected to move to higher frequency bands, causing transmission conditions to be harsher. There are many network solutions available in special or harsh environments. While many of these solutions may be good technical solutions, they are often not cost-effective, resulting in limited deployments. One of the reasons is because they are specially designed to cope with harsh environments. This differentiation in design makes these solutions stand out in performance but lack volumes for economical deployments. The preferred approach should overcome harsh environments while also being suitable for common deployments. In this article, a solution [1] is examined that integrates a proximity network with a radio access network (P-RAN). With this integration, P-RAN has the potential to provide a cost-effective solution not only for indoor areas, shopping malls, and sport arenas, but also for normal macrocell environments.

Istanbul as a "City of Design"

Istanbul as a "City of Design"

Deep solution operators for variational inequalities via proximal neural networks

Following Bauschke and Combettes (Convex analysis and monotone operator theory in Hilbert spaces, Springer, Cham, 2017), we introduce ProxNet, a collection of deep neural networks with ReLU activation which emulate numerical solution operators of variational inequalities (VIs). We analyze the expression rates of ProxNets in emulating solution operators for variational inequality problems posed on closed, convex cones in real, separable Hilbert spaces, covering the classical contact problems in mechanics, and early exercise problems as arise, e.g., in valuation of American-style contracts in Black–Scholes financial market models. In the finite-dimensional setting, the VIs reduce to matrix VIs in Euclidean space, and ProxNets emulate classical projected matrix iterations, such as projected Jacobi and projected SOR methods.

A Novel Spatiotemporal Behavior-Enabled Random Walk Strategy on Online Social Platforms

Location-Based Social Networks have been widely studied in recent years; new approaches constantly developed to solve individuals’ trajectory prediction tasks. However, most of these methods require sufficient data to learn individual features, which is not always satisfied in real situations, especially for online data. The digital data on human behavior typically follows a power-law distribution, indicating that only a few people have rich activities recorded while most people’s behavioral data are limited. In order to overcome this hurdle, our work constructs the user behavior proximity network (UBPN) and proposes a new walking strategy based on this network that extracts the hidden information from the social contacts to substitute the unobserved behavioral information of an individual. Specifically, our proposed walking strategy has two walking paths, accounting for the temporal and social information on the ego users’ and their alters’ mobility activities. This walking strategy is model-agnostic and can be integrated with many existing walk-based deep learning methods. Our work applies the methods on two real-world datasets with rich spatiotemporal information and shows that the performances of the existing prediction methods improve significantly by integrating the proposed walking strategy.

Identifying Candidate Flavonoids for Non-Alcoholic Fatty Liver Disease by Network-Based Strategy.

Nonalcoholic fatty liver disease (NAFLD) is the most common type of chronic liver disease and lacks guaranteed pharmacological therapeutic options. In this study, we applied a network-based framework for comprehensively identifying candidate flavonoids for the prevention and/or treatment of NAFLD. Flavonoid-target interaction information was obtained from combining experimentally validated data and results obtained using a recently developed machine-learning model, AI-DTI. Flavonoids were then prioritized by calculating the network proximity between flavonoid targets and NAFLD-associated proteins. The preventive effects of the candidate flavonoids were evaluated using FFA-induced hepatic steatosis in HepG2 and AML12 cells. We reconstructed the flavonoid-target network and found that the number of re-covered compound-target interactions was significantly higher than the chance level. Proximity scores have successfully rediscovered flavonoids and their potential mechanisms that are reported to have therapeutic effects on NAFLD. Finally, we revealed that discovered candidates, particularly glycitin, significantly attenuated lipid accumulation and moderately inhibited intracellular reactive oxygen species production. We further confirmed the affinity of glycitin with the predicted target using molecular docking and found that glycitin targets are closely related to several proteins involved in lipid metabolism, inflammatory responses, and oxidative stress. The predicted network-level effects were validated at the levels of mRNA. In summary, our study offers and validates network-based methods for the identification of candidate flavonoids for NAFLD.

P336 ELECTRONIC HEALTH RECORDS

IntroductionThe crisis connected with Covid–19 has had strong economic and social repercussions. However, during the state of emergency with restricted access to healthcare, healthcare institutions and users have tested the potential of dematerialization of services thanks to digitalization. The PNRR have maintained in the health sector (Mission6 Health) a fixed point on the horizontal strategic axis of digitalization and innovation, focusing on proximity networks, intermediate structures, telemedicine for territorial health care (Community Homes and Community Hospitals) and on the research and digitalization of the NHS (Electronic Health Record and Essential Levels of Assistance).Our proposal identifies an integrated path that, with the active participation of professionals from the Hospital, the Territory, GPs, social services and families, it allows an appropriate and rapid taking in charge of the patient in hospital dischargeMethodsFrom the SDO office with the Quany we found some indicators of Intensive Cardiology: average weight of ordinary DRG (1.8vs2), average hospital stay (9.05vs9.42), rotation index of beds (51vs39) and pertinence of hospitalizations for acute care / mean hospitalization (9.48 vs 9.38) respectively of 2019–2020 in order to be able to identify critical issues and solutions in the pathways of the frail and improve performance.We used the BPI technique by implementing the Electronic Medical Record and integrating modules that can facilitate a protected discharge by reducing the fragmentation of assistance between the different healthcare structures.We have evaluated the medical–legal questions and identified indicators (average time between arrival of the request and evaluation by the figures involved in the Protected Discharge, time average between evaluation and admission to RSA / Intermediate Care, number of patients discharged with treatment on territory.Expected ResultsThe mapping of the revised process by sharing informations (patient‘s condition and social context), determines the most appropriate post–discharge intervention setting while containing costs.Conclusion The process of taking charge of the territory allowing for benchmarking and control of the quality levels of the care provided.

Interactome of SARS-CoV-2 Modulated Host Proteins With Computationally Predicted PPIs: Insights From Translational Systems Biology Studies

Accelerated efforts to identify intervention strategies for the COVID-19 pandemic caused by SARS-CoV-2 need to be supported by deeper investigations into host invasion and response mechanisms. We constructed the neighborhood interactome network of the 332 human proteins targeted by SARS-CoV-2 proteins, augmenting it with 1,941 novel human protein-protein interactions predicted using our High-precision Protein-Protein Interaction Prediction (HiPPIP) model. Novel interactors, and the interactome as a whole, showed significant enrichment for genes differentially expressed in SARS-CoV-2-infected A549 and Calu-3 cells, postmortem lung samples of COVID-19 patients and blood samples of COVID-19 patients with severe clinical outcomes. The PPIs connected host proteins to COVID-19 blood biomarkers, ACE2 (SARS-CoV-2 entry receptor), genes differentiating SARS-CoV-2 infection from other respiratory virus infections, and SARS-CoV-targeted host proteins. Novel PPIs facilitated identification of the cilium organization functional module; we deduced the potential antiviral role of an interaction between the virus-targeted NUP98 and the cilia-associated CHMP5. Functional enrichment analyses revealed promyelocytic leukaemia bodies, midbody, cell cycle checkpoints and tristetraprolin pathway as potential viral targets. Network proximity of diabetes and hypertension associated genes to host proteins indicated a mechanistic basis for these co-morbidities in critically ill/non-surviving patients. Twenty-four drugs were identified using comparative transcriptome analysis, which include those undergoing COVID-19 clinical trials, showing broad-spectrum antiviral properties or proven activity against SARS-CoV-2 or SARS-CoV/MERS-CoV in cell-based assays. The interactome is available on a webserver at http://severus.dbmi.pitt.edu/corona/.

Protein proximity networks and functional evaluation of the casein kinase 1 gamma family reveal unique roles for CK1γ3 in WNT signaling

Aberrant activation or suppression of WNT/β-catenin signaling contributes to cancer initiation and progression, neurodegeneration, and bone disease. However, despite great need and more than 40 years of research, targeted therapies for the WNT pathway have yet to be fully realized. Kinases are considered exceptionally druggable and occupy key nodes within the WNT signaling network, but several pathway-relevant kinases remain understudied and “dark.” Here, we studied the function of the casein kinase 1γ (CSNK1γ) subfamily of human kinases and their roles in WNT signaling. miniTurbo-based proximity biotinylation and mass spectrometry analysis of CSNK1γ1, CSNK1γ2, and CSNK1γ3 revealed numerous components of the β-catenin–dependent and β-catenin–independent WNT pathways. In gain-of-function experiments, we found that CSNK1γ3 but not CSNK1γ1 or CSNK1γ2 activated β-catenin–dependent WNT signaling, with minimal effect on other signaling pathways. We also show that within the family, CSNK1γ3 expression uniquely induced low-density lipoprotein receptor–related protein 6 phosphorylation, which mediates downstream WNT signaling transduction. Conversely, siRNA-mediated silencing of CSNK1γ3 alone had no impact on WNT signaling, though cosilencing of all three family members decreased WNT pathway activity. Finally, we characterized two moderately selective and potent small-molecule inhibitors of the CSNK1γ family. We show that these inhibitors and a CSNK1γ3 kinase–dead mutant suppressed but did not eliminate WNT-driven low-density lipoprotein receptor–related protein 6 phosphorylation and β-catenin stabilization. Our data suggest that while CSNK1γ3 expression uniquely drives pathway activity, potential functional redundancy within the family necessitates loss of all three family members to suppress the WNT signaling pathway.

Inertial stochastic PALM and applications in machine learning

Inertial algorithms for minimizing nonsmooth and nonconvex functions as the inertial proximal alternating linearized minimization algorithm (iPALM) have demonstrated their superiority with respect to computation time over their non inertial variants. In many problems in imaging and machine learning, the objective functions have a special form involving huge data which encourage the application of stochastic algorithms. While algorithms based on stochastic gradient descent are still used in the majority of applications, recently also stochastic algorithms for minimizing nonsmooth and nonconvex functions were proposed. In this paper, we derive an inertial variant of a stochastic PALM algorithm with variance-reduced gradient estimator, called iSPALM, and prove linear convergence of the algorithm under certain assumptions. Our inertial approach can be seen as generalization of momentum methods widely used to speed up and stabilize optimization algorithms, in particular in machine learning, to nonsmooth problems. Numerical experiments for learning the weights of a so-called proximal neural network and the parameters of Student-t mixture models show that our new algorithm outperforms both stochastic PALM and its deterministic counterparts.

Proximal Instance Aggregator networks for explainable security vulnerability detection

Security vulnerabilities in software are the root cause of cyberattacks. Considering that these defects have huge associated costs, they should be proactively detected and resolved before shipping the software. Data-driven approaches like Artificial Intelligence (AI) are vastly explored for automatic vulnerability detection, given their potential to leverage large-scale vulnerability data feeds and learn from these scenarios. This work introduces a novel Proximal Instance Aggregator (PIA) neural network for accurately capturing insecure C code patterns from Abstract Syntax Tree (AST). It is built upon the concept of Multiple Instance Learning (MIL), which treats the AST representation of the code as a ‘bag’ of tree path ‘instances’. The security vulnerability can manifest in one or multiple such AST path instances. The PIA model dynamically learns a set of abstract concepts to describe the patterns associated with the AST paths. Specifically, the vulnerable nature of an AST path is characterized by its proximity to these concepts. The model also employs the attention mechanism to generate deep representations. By drawing cross-correlation of features between the path instances, the self-attention robustly weighs the relevance of each AST path towards vulnerability classification. The MIL utilizes these deep feature sets to construct the concept space. Thus, even without explicit supervision for localizing the line of defect, the AI automatically learns AST instance classification in a weakly supervised manner. Since AST-level prediction is formed as an aggregation of instance classifications, the AI is inherently explainable. The model outperforms state-of-the-art methods by a fair margin. It achieves 95.63% detection accuracy and 95.65% F1-score on the benchmarked NIST SARD, NVD datasets for a range of vulnerabilities.

Network pharmacology-based approach to evaluate the effect and predict the mechanism of Calculus bovis against oral ulcers

Calculus bovis (CB), a famous traditional Chinese medicine, is commonly used to treat various inflammatory diseases, including oral ulcers. However, the potential mechanisms of CB in the treatment of oral ulcers are still limited. The present study aimed to use a network pharmacology-based approach to understand the multi-compound, multi-target, multi-pathway effect of CB against oral ulcers. The network pharmacology included the following main parts: first, we screened bioactive compounds and predicted corresponding targets via multiple databases; second, we used R software to analyze related Gene Ontology (GO) and KEGG terms; third, we calculated the network proximity on the protein–protein interaction background and identified potential modules. Finally, a rat oral ulcer model was used to validate the therapeutic effect of CB against oral ulcers and test the expression of important targets. Through network pharmacology analysis, 12 bioactive compounds were initially screened out. The network proximity Z score was -4.37, indicating the potential therapeutic efficacy of CB against oral ulcers. Six potential modules were identified, while module 1 was related to the NOD-like receptor signaling pathway and module 6 was related to apoptosis. Experiments confirmed the therapeutic efficacy of CB against oral ulcers. Western blotting indicated that CB upregulated the anti-apoptosis protein Bcl-xl and downregulated the pro-apoptosis protein Bax and the nlrp3 signal. In conclusion, CB exerted a curative effect on oral ulcers by inhibiting apoptosis and nlrp3 activation.

LIGHTWEIGHT MOBILE WEB SERVICE PROVISIONING FOR THE INTERNET OF THINGS MEDIATION

Emerging sensor-embedded smartphones motivated the mobile Internet of Things research. With the integrated embedded hardware and software sensor components, and mobile network technologies, smartphones are capable of providing various environmental context information via embedded mobile device-hosted Web services (MWS). MWS enhances the capability of various mobile sensing applications such as mobile crowdsensing, real time mobile health monitoring, mobile social network in proximity and so on. Although recent smartphones are quite capable in terms of mobile data transmission speed and computation power, the frequent usage of high performance multi-core mobile CPU and the high speed 3G/4G mobile Internet data transmission will quickly drain the battery power of the mobile device. Although numerous previous researchers have tried to overcome the resource intensive issues in mobile embedded service provisioning domain, most of the efforts were constrained because of the underlying resource intensive technologies. This paper presents a lightweight mobile Web service provisioning framework for mobile sensing which utilises the protocols that were designed for constrained Internet of Things environment. The prototype experimental results show that the proposed framework can provide higher throughput and less resource consumption than the traditional mobile Web service frameworks.

Effect of a two-dose vs three-dose vaccine strategy in residential colleges using an empirical proximity network

When SARS-CoV-2 escalated to a pandemic in early 2020, universities and colleges were forced to pivot to virtual instruction. As of fall 2021, many institutions of higher education had reopened, adopting and often mandating mitigation measures like indoor masking, social distancing, regular testing, and vaccination. However, the emergence of the Omicron B.1.1.529 variant in November 2021 disrupted this new routine. The Omicron variant has been shown to be highly transmissible, with researchers estimating the effective reproduction number of Omicron as 3.19 times higher than that of the Delta strain (Ito et al., 2021).

Novel Protein-Protein Interactions Highlighting the Crosstalk between Hypoplastic Left Heart Syndrome, Ciliopathies and Neurodevelopmental Delays.

Hypoplastic left heart syndrome (HLHS) is a severe congenital heart disease (CHD) affecting 1 in 5000 newborns. We constructed the interactome of 74 HLHS-associated genes identified from a large-scale mouse mutagenesis screen, augmenting it with 408 novel protein–protein interactions (PPIs) using our High-Precision Protein–Protein Interaction Prediction (HiPPIP) model. The interactome is available on a webserver with advanced search capabilities. A total of 364 genes including 73 novel interactors were differentially regulated in tissue/iPSC-derived cardiomyocytes of HLHS patients. Novel PPIs facilitated the identification of TOR signaling and endoplasmic reticulum stress modules. We found that 60.5% of the interactome consisted of housekeeping genes that may harbor large-effect mutations and drive HLHS etiology but show limited transmission. Network proximity of diabetes, Alzheimer’s disease, and liver carcinoma-associated genes to HLHS genes suggested a mechanistic basis for their comorbidity with HLHS. Interactome genes showed tissue-specificity for sites of extracardiac anomalies (placenta, liver and brain). The HLHS interactome shared significant overlaps with the interactomes of ciliopathy- and microcephaly-associated genes, with the shared genes enriched for genes involved in intellectual disability and/or developmental delay, and neuronal death pathways, respectively. This supported the increased burden of ciliopathy variants and prevalence of neurological abnormalities observed among HLHS patients with developmental delay and microcephaly, respectively.

Analysis on the characteristics of spatio-temporal evolution and aggregation trend of early COVID-19 in mainland China

To analyze the spatio-temporal aggregation of COVID-19 in mainland China within 20 days after the closure of Wuhan city, and provide a theoretical basis for formulating scientific prevention measures in similar major public health events in the future. Draw a distribution map of the cumulative number of COVID-19 by inverse distance weighted interpolation; analyze the spatio-temporal characteristics of the daily number of COVID-19 in mainland China by spatio-temporal autocorrelation analysis; use the spatio-temporal scanning statistics to detect the spatio-temporal clustering area of the daily number of new diagnosed cases. The cumulative number of diagnosed cases obeyed the characteristics of geographical proximity and network proximity to Hubei. Hubei and its neighboring provinces were most affected, and the impact in the eastern China was more dramatic than the impact in the western; the global spatio-temporal Moran’s I index showed an overall downward trend. Since the 10th day of the closure of Wuhan, the epidemic in China had been under effective control, and more provinces had shifted into low-incidence areas. The number of new diagnosed cases had gradually decreased, showing a random distribution in time and space (P< 0.1), and no clusters were formed. Conclusion: the spread of COVID-19 had obvious spatial-temporal aggregation. China’s experience shows that isolation city strategy can greatly contain the spread of the COVID-19 epidemic.

Prediction of Synergistic Antibiotic Combinations by Graph Learning.

Antibiotic resistance is a major public health concern. Antibiotic combinations, offering better efficacy at lower doses, are a useful way to handle this problem. However, it is difficult for us to find effective antibiotic combinations in the vast chemical space. Herein, we propose a graph learning framework to predict synergistic antibiotic combinations. In this model, a network proximity method combined with network propagation was used to quantify the relationships of drug pairs, and we found that synergistic antibiotic combinations tend to have smaller network proximity. Therefore, network proximity can be used for building an affinity matrix. Subsequently, the affinity matrix was fed into a graph regularization model to predict potential synergistic antibiotic combinations. Compared with existing methods, our model shows a better performance in the prediction of synergistic antibiotic combinations and interpretability.

Impact of Grocery Store Proximity on Store Preference Among Atlanta SNAP-Ed Participants

ObjectiveTo assess the association between grocery store proximities and the individual's grocery store preferences among Supplemental Nutrition Assistance Program Education participants in Atlanta. MethodsUniversity of Georgia Supplemental Nutrition Assistance Program Education participants (n = 615, response rate is 36%) in 3 counties provided their preferred grocery store chains. The association between store proximity (both network distance and driving time) and store preference was measured through logistic regression controlling for age, sex, and race. ResultsDescriptive statistics showed participants had widely varying proximities to grocery stores. Model results were significant for all smaller chains (Aldi, Big Bear, Wayfield, Food Depot, and Save-A-Lot), Kroger (P < 0.01), as well as for Walmart (time only, P = 0.002). Conclusions and ImplicationsFuture studies might identify whether local groceries are more willing to partner on interventions or are more effective at reaching local residents. Surveys or techniques such as sketch mapping could also show whether individuals shop in neighborhoods close to work or friends and family.

Abstract EP12: Uric Acid Predicts Recovery Of Left Ventricular Function And Adverse Events In Heart Failure With Reduced Ejection Fraction: Potential Mechanistic Insight From Network Analyses

Background and Aims: Heart failure with reduced ejection fraction (HFrEF) still carries a high risk for a sustained decrease in left ventricular ejection fraction (LVEF) even with the optimal medical therapy. Currently, there is no effective tool to stratify these patients according to their recovery potential. We tested the hypothesis that uric acid (UA) could predict recovery of LVEF and prognosis of HFrEF patients and attempted to explore mechanistic relationship between hyperuricemia and HFrEF. Methods: HFrEF patients with hyperuricemia were selected from the National Inpatient Sample (NIS) 2016-2018 database and our Xianyang prospective cohort study. Demographics, cardiac risk factors, and cardiovascular events were identified. Network-based analysis was utilized to examine the relationship between recovery of LVEF and hyperuricemia, and we further elucidated the underlying mechanisms for the impact of hyperuricemia on HFrEF. Results: After adjusting confounding factors by propensity score matching, hyperuricemia was a determinant of HFrEF [OR 1.247 (1.172-1.328); P &lt;0.001] of NIS dataset. In Xianyang prospective cohort study, hyperuricemia is a significant and independent risk factor for all-cause death (adjusted HR 2.387, 95% CI 1.141-4.993; P =0.021), heart failure readmission (adjusted HR 1.848, 95% CI 1.048-3.259; P =0.034), and composite events (adjusted HR 1.706, 95% CI 1.001-2.906; P =0.049) in HFrEF patients. UA value at baseline was negatively correlated to LVEF of follow-ups (r=-0.19; P =0.046). Cutoff UA value of 312.5 μmmol/L at baseline can work as a predictor of LVEF recovery during follow-up, with the sensitivity of 66.7%, the specificity of 35.1%, and the accuracy of 0.668 (95% CI, 0.561-0.775; P =0.006). Moreover, gene overlap analysis and network proximity analysis demonstrated a strong correlation between HFrEF and Hyperuricemia. Conclusion: Lower baseline UA value predicted the LVEF recovery and less long-term adverse events in HFrEF patients. Our results provide new insights into underlying mechanistic relationship between hyperuricemia and HFrEF.

Prediction of hospital-onset COVID-19 using networks of patient contact: an observational study

PurposePredicting healthcare-acquired infections (HAIs) has the potential to revolutionise the prevention and control of transmissible infections. Existing prediction models for HAIs, however, fail to capture fully the contact-driven nature of infectious diseases. Here, we investigate the epidemiological predictivity of patient contact patterns through a forecasting model for hospital-onset COVID-19 infection (HOCI).Methods & MaterialsOur cohort comprises all patient admissions at a large London NHS Trust between 1/04/2020 and 1/04/2021. For patients, we consider (i) their hospital pathway, (ii) patient contacts, and (iii) date of COVID-19 infection. We consider rolling 14-day windows and forecast patient infection over the subsequent 7 days. Over each window, we construct a patient contact network and compute network features that capture contact centrality. We then combine network features, hospital environmental variables and patient clinical data to predict subsequent infections.ResultsA total of 51,157 patient admissions/episodes were observed during the study. Across all models, we find that contact-network features showed the highest performance (0.91 AUC-ROC). A reduced model with the six most predictive variables was almost as predictive and contained five features from patient contact (including direct contact with and network proximity to infectious cases) and only one environmental variable (length of stay).ConclusionOur results reveal that the number of direct contacts and network proximity to infectious patient(s) are highly predictive of HOCI. Such contact-based risk factors are easily extracted from routinely collected electronic health records providing a highly accessible route to improve personalised disease prognostics in future models.

Sex-specific responses to anthropogenic risk shape wild chimpanzee social networks in a human-impacted landscape

Large animal species are most likely to survive human-induced rapid environmental change if they display high levels of behavioural flexibility. Examining social responses in species that form closely bonded social groups and display high fission–fusion dynamics, such as chimpanzees, Pan troglodytes , will help determine their resilience to dynamic human activities and the potential for sustainable human–wildlife interactions. Coinciding with the seasonal availability of cultivated jackfruit, Artocarpus heterophyllus , a dominant food available in croplands and village gardens, we examined the social adjustments of wild chimpanzees to risks presented by a human-modified landscape at Bulindi, Uganda. Absolute party size decreased when chimpanzees moved from low-risk ‘natural’ habitat (forest fragments) to high-risk ‘anthropogenic’ habitat (croplands and other village areas), driven mainly by parties containing fewer females. Using social network analysis, we found that chimpanzee social structure showed partial flexibility in response to anthropogenic risk. The strength of social ties in a 5 m proximity network stayed similar and individual centrality measures were correlated between low- and high-risk habitats. Nevertheless, in high-risk habitat males were more central than females and exclusively dominated the core of the network's core–periphery structure. However, sex differences were not apparent when we corrected for party membership, indicating that differences were driven by changes to party membership between habitat types rather than changes in social proximity within parties. We suggest that perception of anthropogenic risk can drive continual adjustments to grouping dynamics in response to changing conditions that modify social lives of male and female chimpanzees in different ways. Consequently, changes in grouping patterns induced by continuing anthropogenic change could impact ecological and evolutionary processes influenced by social structure, including cultural evolution and the spread of infectious disease. • Party size decreased from low-risk ‘natural’ to high-risk ‘anthropogenic’ habitat. • Males had higher social centrality than females in high-risk habitat only. • Sex differences in centrality were caused by females avoiding high-risk habitat. • No sex difference in centrality was apparent when controlling for party membership. • Chimpanzee social structure showed partial flexibility to anthropogenic risk.