Graph Of Function Research Articles

Prognostic Role of Hemispherical Functional Connectivity in Stroke: A Study via Graph Theory Versus Coherence of Electroencephalography Rhythms.

The objective of the present study is to explore whether acute stroke may result in changes in brain network architecture by electroencephalography functional coupling analysis and graph theory. Ninety acute stroke patients and 110 healthy subjects were enrolled in different clinical centers in Rome, Italy, starting from 2013, and for each one electroencephalographies were recorded within <15 days from stroke onset. All patients were clinically evaluated through National Institutes of Health Stroke Scale, Barthel Index, and Action Research Arm Test in the acute stage and during the follow-up. Functional connectivity was assessed using Total Coherence and Small World (SW) by comparing the affected and the unaffected hemisphere between groups (Stroke versus Healthy). Correlations between connectivity and poststroke recovery scores have been carried out. In stroke patients, network hemispheric asymmetry, in terms of Total Coherence, was mainly detected in the affected hemisphere with lower values in Delta, Theta, Alpha1, and Alpha2 (P=0.000001), whereas the unaffected hemisphere showed lower Total Coherence only in Delta and Theta (P=0.000001). SW revealed a significant difference only in the affected hemisphere in all electroencephalography bands (lower SW in Delta (P=0.000003), Theta (P=0.000003), Alpha1 (P=0.000203), and Alpha2 (P=0.028) and higher SW in Beta2 (P=0.000002) and Gamma (P=0.000002)). We also found significant correlations between SW and improvement in National Institutes of Health Stroke Scale (Theta SW: r=-0.2808), Barthel Index (Delta SW: r=0.3692; Theta SW: r=0.3844, Beta2 SW: r=-0.3589; Gamma SW: r=-04948), and Action Research Arm Test (Beta2 SW: r=-0.4274; Gamma SW: r=-0.4370). These findings demonstrated changes in global functional connectivity and in the balance of network segregation and integration induced by acute stroke. The findings on the correlations between clinical outcome(s) and poststroke network architecture indicate the possibility to identify a predictive index of recovery useful to address and personalize the rehabilitation program.

MRI biomarkers of freezing of gait development in Parkinson’s disease

This study investigated longitudinal clinical, structural and functional brain alterations in Parkinson’s disease patients with freezing of gait (PD-FoG) and in those developing (PD-FoG-converters) and not developing FoG (PD-non-converters) over two years. Moreover, this study explored if any clinical and/or MRI metric predicts FoG development. Thirty PD-FoG, 11 PD-FoG-converters and 11 PD-non-converters were followed for two years. Thirty healthy controls were included at baseline. Participants underwent clinical and MRI visits. Cortical thickness, basal ganglia volumes and functional network graph metrics were evaluated at baseline and over time. In PD groups, correlations between baseline MRI and clinical worsening were tested. A ROC curve analysis investigated if baseline clinical and MRI measures, selected using a stepwise model procedure, could differentiate PD-FoG-converters from PD-non-converters. At baseline, PD-FoG patients had widespread cortical/subcortical atrophy, while PD-FoG-converters and non-converters showed atrophy in sensorimotor areas and basal ganglia relative to controls. Over time, PD-non-converters accumulated cortical thinning of left temporal pole and pallidum without significant clinical changes. PD-FoG-converters showed worsening of disease severity, executive functions, and mood together with an accumulation of occipital atrophy, similarly to PD-FoG. At baseline, PD-FoG-converters relative to controls and PD-FoG showed higher global and parietal clustering coefficient and global local efficiency. Over time, PD-FoG-converters showed reduced parietal clustering coefficient and sensorimotor local efficiency, PD-non-converters showed increased sensorimotor path length, while PD-FoG patients showed stable graph metrics. Stepwise prediction model including dyskinesia, postural instability and gait disorders scores and parietal clustering coefficient was the best predictor of FoG conversion. Combining clinical and MRI data, ROC curves provided the highest classification power to predict the conversion (AUC = 0.95, 95%CI: 0.86–1). Structural MRI is a useful tool to monitor PD progression, while functional MRI together with clinical features may be helpful to identify FoG conversion early.

Almost continuous Sierpiński–Zygmund functions under different set-theoretical assumptions

A function f:mathbb {R}rightarrow mathbb {R} is: almost continuous in the sense of Stallings, fin textrm{AC}, if each open set Gsubset mathbb {R}^2 containing the graph of f contains also the graph of a continuous function g:mathbb {R}rightarrow mathbb {R}; Sierpiński–Zygmund, fin textrm{SZ} (or, more generally, fin textrm{SZ}(textrm{Bor})), provided its restriction frestriction M is discontinuous (not Borel, respectively) for any Msubset mathbb {R} of cardinality continuum. It is known that an example of a Sierpiński–Zygmund almost continuous function f:mathbb {R}rightarrow mathbb {R} (i.e., an fin textrm{SZ}cap textrm{AC}) cannot be constructed in ZFC; however, an fin textrm{SZ}cap textrm{AC} exists under the additional set-theoretical assumption {{,textrm{cov},}}(mathcal {M})=mathfrak {c}, that is, that mathbb {R} cannot be covered by less than mathfrak {c}-many meager sets. The primary purpose of this paper is to show that the existence of an fin textrm{SZ}cap textrm{AC} is also consistent with ZFC plus the negation of {{,textrm{cov},}}(mathcal {M})=mathfrak {c}. More precisely, we show that it is consistent with ZFC+{{,textrm{cov},}}(mathcal {M})<mathfrak {c} (follows from the assumption that {{,textrm{non},}}(mathcal {N})<{{,textrm{cov},}}(mathcal {N})=mathfrak {c}) that there is an fin textrm{SZ}(textrm{Bor})cap textrm{AC} and that such a map may have even stronger properties expressed in the language of Darboux-like functions. We also examine, assuming either {{,textrm{cov},}}(mathcal {M})=mathfrak {c} or {{,textrm{non},}}(mathcal {N})<{{,textrm{cov},}}(mathcal {N})=mathfrak {c}, the lineability and the additivity coefficient of the class of all almost continuous Sierpiński–Zygmund functions. Several open problems are also stated.

Diagnosis of Epilepsy with Functional Connectivity in EEG after a Suspected First Seizure.

Epilepsy is regarded as a structural and functional network disorder, affecting around 50 million people worldwide. A correct disease diagnosis can lead to quicker medical action, preventing adverse effects. This paper reports the design of a classifier for epilepsy diagnosis in patients after a first ictal episode, using electroencephalogram (EEG) recordings. The dataset consists of resting-state EEG from 629 patients, of which 504 were retained for the study. The patient's cohort exists out of 291 patients with epilepsy and 213 patients with other pathologies. The data were split into two sets: 80% training set and 20% test set. The extracted features from EEG included functional connectivity measures, graph measures, band powers and brain asymmetry ratios. Feature reduction was performed, and the models were trained using Machine Learning (ML) techniques. The models' evaluation was performed with the area under the receiver operating characteristic curve (AUC). When focusing specifically on focal lesional epileptic patients, better results were obtained. This classification task was optimized using a 5-fold cross-validation, where SVM using PCA for feature reduction achieved an AUC of 0.730 ± 0.030. In the test set, the same model achieved 0.649 of AUC. The verified decrease is justified by the considerable diversity of pathologies in the cohort. An analysis of the selected features across tested models shows that functional connectivity and its graph measures have the most considerable predictive power, along with full-spectrum frequency-based features. To conclude, the proposed algorithms, with some refinement, can be of added value for doctors diagnosing epilepsy from EEG recordings after a suspected first seizure.

Novel favipiravir pattern-based learning model for automated detection of specific language impairment disorder using vowels.

Specific language impairment (SLI) is one of the most common diseases in children, and early diagnosis can help to obtain better timely therapy economically. It is difficult and time-consuming for clinicians to accurately detect SLI through standard clinical assessments. Hence, machine learning algorithms have been developed to assist in the accurate diagnosis of SLI. This work aims to investigate the graph of the favipiravir molecule-based feature extraction function and propose an accurate SLI detection model using vowels. We proposed a novel handcrafted machine learning framework. This architecture comprises the favipiravir molecular structure pattern, statistical feature extractor, wavelet packet decomposition (WPD), iterative neighborhood component analysis (INCA), and support vector machine (SVM) classifier. Two feature extraction models, statistical and textural, are employed in the handcrafted feature generation methodology. A new nature-inspired graph-based feature extractor that uses the chemical depiction of the favipiravir (favipiravir became popular with the COVID-19 pandemic) is employed for feature extraction. Finally, the proposed favipiravir pattern, statistical feature extractor, and wavelet packet decomposition are used to create a feature vector. Moreover, a statistical feature extractor is used in this work. The WPD generates multilevel features, and the most meaningful features are selected using the NCA feature selector. Finally, these chosen features are fed to SVM classifier for automated classification. Two validation methods, (i) leave one subject out (LOSO) and (ii) tenfold cross-validations (CV), are used to obtain robust classification results. Our proposed favipiravir pattern-based model developed using a vowel dataset can detect SLI children with an accuracy of 99.87% and 98.86% using tenfold and LOSO CV strategies, respectively. These results demonstrated the high vowel classification ability of the proposed favipiravir pattern-based model.

Graph-generative neural network for EEG-based epileptic seizure detection via discovery of dynamic brain functional connectivity

Dynamic complexity in brain functional connectivity has hindered the effective use of signal processing or machine learning methods to diagnose neurological disorders such as epilepsy. This paper proposed a new graph-generative neural network (GGN) model for the dynamic discovery of brain functional connectivity via deep analysis of scalp electroencephalogram (EEG) signals recorded from various regions of a patient’s scalp. Brain functional connectivity graphs are generated for the extraction of spatial–temporal resolution of various onset epilepsy seizure patterns. Our supervised GGN model was substantiated by seizure detection and classification experiments. We train the GGN model using a clinically proven dataset of over 3047 epileptic seizure cases. The GGN model achieved a 91% accuracy in classifying seven types of epileptic seizure attacks, which outperformed the 65%, 74%, and 82% accuracy in using the convolutional neural network (CNN), graph neural networks (GNN), and transformer models, respectively. We present the GGN model architecture and operational steps to assist neuroscientists or brain specialists in using dynamic functional connectivity information to detect neurological disorders. Furthermore, we suggest to merge our spatial–temporal graph generator design in upgrading the conventional CNN and GNN models with dynamic convolutional kernels for accuracy enhancement.

Visual Saliency and Image Reconstruction from EEG Signals via an Effective Geometric Deep Network-Based Generative Adversarial Network

Reaching out the function of the brain in perceiving input data from the outside world is one of the great targets of neuroscience. Neural decoding helps us to model the connection between brain activities and the visual stimulation. The reconstruction of images from brain activity can be achieved through this modelling. Recent studies have shown that brain activity is impressed by visual saliency, the important parts of an image stimuli. In this paper, a deep model is proposed to reconstruct the image stimuli from electroencephalogram (EEG) recordings via visual saliency. To this end, the proposed geometric deep network-based generative adversarial network (GDN-GAN) is trained to map the EEG signals to the visual saliency maps corresponding to each image. The first part of the proposed GDN-GAN consists of Chebyshev graph convolutional layers. The input of the GDN part of the proposed network is the functional connectivity-based graph representation of the EEG channels. The output of the GDN is imposed to the GAN part of the proposed network to reconstruct the image saliency. The proposed GDN-GAN is trained using the Google Colaboratory Pro platform. The saliency metrics validate the viability and efficiency of the proposed saliency reconstruction network. The weights of the trained network are used as initial weights to reconstruct the grayscale image stimuli. The proposed network realizes the image reconstruction from EEG signals.

Functional network properties derived from wide-field calcium imaging differ with wakefulness and across cell type

To improve ‘bench-to-bedside’ translation, it is integral that knowledge flows bidirectionally—from animal models to humans, and vice versa. This requires common analytical frameworks, as well as open software and data sharing practices. We share a new pipeline (and test dataset) for the preprocessing of wide-field optical fluorescence imaging data—an emerging mode applicable in animal models—as well as results from a functional connectivity and graph theory analysis inspired by recent work in the human neuroimaging field. The approach is demonstrated using a dataset comprised of two test-cases: (1) data from animals imaged during awake and anesthetized conditions with excitatory neurons labeled, and (2) data from awake animals with different genetically encoded fluorescent labels that target either excitatory neurons or inhibitory interneuron subtypes. Both seed-based connectivity and graph theory measures (global efficiency, transitivity, modularity, and characteristic path-length) are shown to be useful in quantifying differences between wakefulness states and cell populations. Wakefulness state and cell type show widespread effects on canonical network connectivity with variable frequency band dependence. Differences between excitatory neurons and inhibitory interneurons are observed, with somatostatin expressing inhibitory interneurons emerging as notably dissimilar from parvalbumin and vasoactive polypeptide expressing cells. In sum, we demonstrate that our pipeline can be used to examine brain state and cell-type differences in mesoscale imaging data, aiding translational neuroscience efforts. In line with open science practices, we freely release the pipeline and data to encourage other efforts in the community.

The thresholding problem and variability in the EEG graph network parameters

Graph thresholding is a frequently used practice of eliminating the weak connections in brain functional connectivity graphs. The main aim of the procedure is to delete the spurious connections in the data. However, the choice of the threshold is arbitrary, and the effect of the threshold choice is not fully understood. Here we present the description of the changes in the global measures of a functional connectivity graph depending on the different proportional thresholds based on the 146 resting-state EEG recordings. The dynamics is presented in five different synchronization measures (wPLI, ImCoh, Coherence, ciPLV, PPC) in sensors and source spaces. The analysis shows significant changes in the graph’s global connectivity measures as a function of the chosen threshold which may influence the outcome of the study. The choice of the threshold could lead to different study conclusions; thus it is necessary to improve the reasoning behind the choice of the different analytic options and consider the adoption of different analytic approaches. We also proposed some ways of improving the procedure of thresholding in functional connectivity research.

Modeling of Climatic Variables using Stochastic Approaches in Sudan

The climatic variables take a significant role in agricultural process and irrigation management because we need to know all changes related to the climate which absolutely will be affected agricultural yield. For this purpose, the ARIMA models were suggested in this study for modeling daily average temperature, relative humidity, and solar radiation variables related to five main meteorological stations (Wad Madani, Khartoum, Al Gadaref, Al Damazin, and Dongola) in Sudan. The daily variables were obtained from the period 2013-to 2020 years. Time series analysis methods are used for estimating and modeling the climatic variables using ARIMA (Autoregressive Integrated Moving Average Models) which are called Box Jenkins models. For modeling purposes, linear stochastic models were used to estimate future values of daily variables. The Augmented Dickey-Fuller test (ADF) was applied to the time series to check the stationarity at 1%, 5%, and 10% confidence levels. The time series of variables were stationarity and without trend. Diagnostic checks were used for checking models and selecting the best models from the autocorrelation (ACF) and partial autocorrelation (PACF) function graphs. The best models were selected according to the adjusted R2, Standard error (S.E), Akaike information criterion (AIC), and Bayesian information criterion (BIC) values. The best results were observed in ARIMA (1,0,1) and (1,0,2) which can be effective for predicting future values. The ARIMA models obtained satisfactory results for temperature, relative humidity, and solar radiation variables. So, this study might be extremely helpful for agricultural engineers to achieve all the processes related to agricultural practices.

Influence of a Function’s Coefficients and Feedback of the Mathematical Work When Reading a Graph in an Online Assessment System

This paper shows the results of an experiment applied to 170 students from two Chilean universities who solve a task about reading a graph of an affine function in an online assessment environment where the parameters (coefficients of the graphed affine function) are randomly defined from an ad-hoc algorithm, with automatic correction and automatic feedback. We distinguish two versions: one of them with integer coefficients and the other one with decimal coefficients in the affine function. We observed that the nature of the coefficients impacts the mathematical work used by the students, where we again focus on two of them: by direct estimation from the graph or by calculating the equation of the line. On the other hand, feedback oriented towards the "estimation" strategy influences the mathematical work used by the students, even though a non-negligible group persists in the "calculating" strategy, which is partly explained by the perception of each of the strategies.

Urban road traffic flow prediction: A graph convolutional network embedded with wavelet decomposition and attention mechanism

Urban road traffic flow prediction is the key basis for the development of Intelligent Transportation System. The complex urban structure leads to irregular shape and layout of the road network, which poses a challenge to capture the spatio-temporal correlation of traffic flow at different nodes in the region. In this study, a graph convolutional network model framework embedded with wavelet decomposition and attention mechanism (WDA-GCN) is proposed to predict the traffic flow of each traffic monitor at the regional level by exploring the spatio-temporal correlation among traffic monitors. Specifically, the spatial correlation between different monitors is encoded into two graphs by Graph Convolutional Network (GCN): geographical neighbor graph and functional similarity graph. The Gated Recurrent Unit (GRU) is used to learn the spatial features extracted by GCN, and the attention mechanism is added to improve the prediction accuracy. Finally, the time series data and spatio-temporal correlation of traffic flow are input into the encoder–decoder based on GRU to realize regional traffic flow prediction. The model is validated and compared with the real traffic monitor data in Daxing District of Beijing, China, and the results show that the prediction accuracy of WDA-GCN model can reach 81.03% after embedding wavelet decomposition and attention mechanism, which is better than the traditional time series prediction methods and deep learning methods.

Disrupted topological organization of resting-state functional brain networks in Parkinson's disease patients with glucocerebrosidase gene mutations.

The mutations of the glucocerebrosidase (GBA) gene are the greatest genetic risk factor for Parkinson's disease (PD). The mechanism underlying the association between GBA mutations and PD has not been fully elucidated. Using resting-state functional magnetic resonance imaging and graph theory analysis to investigate the disrupted topological organization in PD patients with GBA mutation (GBA-PD). Eleven GBA-PD patients, 11 noncarriers with PD, and 18 healthy controls (HCs) with a similar age and sex distribution were recruited. Individual whole-brain functional connectome was constructed, and the global and nodal topological disruptions were calculated among groups. Partial correlation analyses between the clinical features of patients with PD and topological alterations were performed. The GBA-PD group showed prominently decreased characteristic path length (Lp) and increased global efficiency (Eg) compared to HCs at the global level; a significantly increased nodal betweenness centrality in the medial prefrontal cortex (mPFC) and precuneus within the default mode network, and precentral gyrus within the sensorimotor network, while a significantly decreased betweenness centrality in nodes within the cingulo-opercular network compared to the noncarrier group at the regional level. The altered nodal betweenness centrality of mPFC was positively correlated with fatigue severity scale scores in all patients with PD. The preliminary pilot study found that GBA-PD patients had a higher functional integration at the global level. The nodal result of the mPFC is congruent with the potential fatigue pathology in PD and is suggestive of a profound effect of GBA mutations on the clinical fatigue in patients with PD.

Vaginal microbiota networks as a mechanistic predictor of aerobic vaginitis.

Aerobic vaginitis (AV) is a complex vaginal dysbiosis that is thought to be caused by the micro-ecological change of the vaginal microbiota. While most studies have focused on how changes in the abundance of individual microbes are associated with the emergence of AV, we still do not have a complete mechanistic atlas of the microbe-AV link. Network modeling is central to understanding the structure and function of any microbial community assembly. By encapsulating the abundance of microbes as nodes and ecological interactions among microbes as edges, microbial networks can reveal how each microbe functions and how one microbe cooperate or compete with other microbes to mediate the dynamics of microbial communities. However, existing approaches can only estimate either the strength of microbe-microbe link or the direction of this link, failing to capture full topological characteristics of a network, especially from high-dimensional microbial data. We combine allometry scaling law and evolutionary game theory to derive a functional graph theory that can characterize bidirectional, signed, and weighted interaction networks from any data domain. We apply our theory to characterize the causal interdependence between microbial interactions and AV. From functional networks arising from different functional modules, we find that, as the only favorable genus from Firmicutes among all identified genera, the role of Lactobacillus in maintaining vaginal microbial symbiosis is enabled by upregulation from other microbes, rather than through any intrinsic capacity. Among Lactobacillus species, the proportion of L. crispatus to L. iners is positively associated with more healthy acid vaginal ecosystems. In a less healthy alkaline ecosystem, L. crispatus establishes a contradictory relationship with other microbes, leading to population decrease relative to L. iners. We identify topological changes of vaginal microbiota networks when the menstrual cycle of women changes from the follicular to luteal phases. Our network tool provides a mechanistic approach to disentangle the internal workings of the microbiota assembly and predict its causal relationships with human diseases including AV.

Assessment of EEG-based functional connectivity in response to haptic delay.

Haptic technologies enable users to physically interact with remote or virtual environments by applying force, vibration, or motion via haptic interfaces. However, the delivery of timely haptic feedback remains a challenge due to the stringent computation and communication requirements associated with haptic data transfer. Haptic delay disrupts the realism of the user experience and interferes with the quality of interaction. Research efforts have been devoted to studying the neural correlates of delayed sensory stimulation to better understand and thus mitigate the impact of delay. However, little is known about the functional neural networks that process haptic delay. This paper investigates the underlying neural networks associated with processing haptic delay in passive and active haptic interactions. Nineteen participants completed a visuo-haptic task using a computer screen and a haptic device while electroencephalography (EEG) data were being recorded. A combined approach based on phase locking value (PLV) functional connectivity and graph theory was used. To assay the effects of haptic delay on functional connectivity, we evaluate a global connectivity property through the small-worldness index and a local connectivity property through the nodal strength index. Results suggest that the brain exhibits significantly different network characteristics when a haptic delay is introduced. Haptic delay caused an increased manifestation of the small-worldness index in the delta and theta bands as well as an increased nodal strength index in the middle central region. Inter-regional connectivity analysis showed that the middle central region was significantly connected to the parietal and occipital regions as a result of haptic delay. These results are expected to indicate the detection of conflicting visuo-haptic information at the middle central region and their respective resolution and integration at the parietal and occipital regions.

Spacelike minimal surfaces which are graphs in [formula omitted

In this work we study a system of two second order nonlinear partial differential equations corresponding to the equations of minimal spacelike graphs in the four-dimensional Minkownski space R14. We obtain a set of entire solutions of these equations, adapting the method of Karl Kommerell used for the analogous problem in R4. In our technique we replace ‘‘graphs of holomorphic functions” by ‘‘graphs of generalized harmonic curves” and then we obtain a class of minimal spacelike surfaces in R14 which are also entire graphs with non-zero Gauss curvature. We then provide two versions of the Bernstein Theorem corresponding for the two types of graphs which can occur in R14. Several explicit examples are given.

Pluripolar hulls and fine holomorphy

Examples by Poletsky and the author and by Zwonek show the existence of nowhere extendable holomorphic functions with the property that the pluripolar hull of their graphs is much larger than the graph of the respective functions, and contains multiple sheets. We will explain this phenomenon by fine analytic continuation of the function over part of a Cantor-type set involved. This gives more information on the hull, and allows for weakening and effectiveness of the conditions in the original examples.

Counting distinct functional graphs from linear finite dynamical systems

Let Fq be the finite field with q elements and, for each positive integer n, let Aq(n) be the number of non isomorphic functional graphs arising from Fq-linear maps T:Fqn→Fqn. In 2013, Bach and Bridy proved that, for every prime power q, we have that 12≤liminfn→+∞log⁡log⁡Aq(n)log⁡n≤limsupn→+∞log⁡log⁡Aq(n)log⁡n≤1. By combining some ideas from linear algebra, combinatorics and number theory, in this paper we provide sharper estimates on the function Aq(n) and, in particular, we prove that limn→+∞⁡log⁡log⁡Aq(n)log⁡n=1 for every prime power q.

Development of a Problematic Khat Use Screening Test (PKUST-17) in Ethiopia: Classical Test Theory and Item Response Theory Analysis

Problematic khat use has been poorly defined and measured. This study aimed to develop a problematic khat use screening tool in the Gurage Community, South-central Ethiopia. We have used a series of methods (systematic review, qualitative study and cognitive interviewing) to generate a pool of items for problematic khat use screening tool. Classical test theory (inter-item, and item-total correlation, intraclass correlation, Cronbachalpha and exploratory factor analysis) and a 2-parameter item response theory statistics were used for initial psychometric evaluation of the scale. Initially, we developed a pool of 50 items. IRT analysis indicated that the item, ’’during the past three months, how often have you experienced irritability when you didn’t chew khat?” showed the highest discrimination (α thresholds =2.44). The item about “during the past three months, to what extent have you failed to do what was normally expected of you because of your use of khat? was the most difficult (first β thresholds =0.25) item, which was the only positive coefficient. The test information function graph depicts that the scale provided a lot of information in the latent trait’s moderate range. We found that employing series of methods is very important for item selection and refinement. We developed a Problematic Khat Use Screening Test with 17 items which will have a good utility for the general population or primary health care setting more than for the clinical population. Future studies should do full-scale validation of the scale.

Non-parametric data-driven background modelling using conditional probabilities

Background modelling is one of the main challenges in particle physics data analysis. Commonly employed strategies include the use of simulated events of the background processes, and the fitting of parametric background models to the observed data. However, reliable simulations are not always available or may be extremely costly to produce. As a result, in many cases uncertainties associated with the accuracy or sample size of the simulation are the limiting factor in the analysis sensitivity. At the same time, parametric models are limited by the a priori unknown functional form and parameter values of the background distribution. These issues become ever more pressing when large datasets become available, as it is already the case at the CERN Large Hadron Collider, and when studying exclusive signatures involving hadronic backgrounds.A widely applicable approach for non-parametric data-driven background modelling is proposed, which addresses these issues for a broad class of searches and measurements. It relies on a relaxed version of the event selection to estimate conditional probability density functions and two different techniques are discussed for its realisation. The first relies on ancestral sampling and uses data from a relaxed event selection to estimate a graph of conditional probability density functions of the variables used in the analysis, while accounting for significant correlations. A background model is then generated from events drawn from this graph, on which the full event selection is applied. In the second, a novel generative adversarial network is trained to estimate the joint probability density function of the variables used in the analysis. The training is performed on a relaxed event selection, which excludes the signal region, and the network is conditioned on a blinding variable. Subsequently, the conditional probability density function is interpolated into the signal region to model the background. The application of each method on a benchmark analysis and on ensemble tests is presented in detail, and the performance is discussed.