Discrete Data Sets Research Articles

Do morphometric data improve phylogenetic reconstruction? A systematic review and assessment

BackgroundIsolating phylogenetic signal from morphological data is crucial for accurately merging fossils into the tree of life and for calibrating molecular dating. However, subjective character definition is a major limitation which can introduce biases that mislead phylogenetic inferences and divergence time estimation. The use of quantitative data, e.g., geometric morphometric (GMM; shape) data can allow for more objective integration of morphological data into phylogenetic inference. This systematic review describes the current state of the field in using continuous morphometric data (e.g., GMM data) for phylogenetic reconstruction and assesses the efficacy of these data compared to discrete characters using the PRISMA-EcoEvo v1.0. reporting guideline, and offers some pathways for approaching this task with GMM data. A comprehensive search string yielded 11,123 phylogenetic studies published in English up to Oct 2023 in the Web of Science database. Title and abstract screening removed 10,975 articles, and full-text screening was performed for 132 articles. Of these, a total of twelve articles met final inclusion criteria and were used for downstream analyses.ResultsPhylogenetic performance was compared between approaches that employed continuous morphometric and discrete morphological data. Overall, the reconstructed phylogenies did not show increased resolution or accuracy (i.e., benchmarked against molecular phylogenies) as continuous data alone or combined with discrete morphological datasets.ConclusionsAn exhaustive search of the literature for existing empirical continuous data resulted in a total of twelve articles for final inclusion following title/abstract, and full-text screening. Our study was performed under a rigorous framework for systematic reviews, which showed that the lack of available comparisons between discrete and continuous data hinders our understanding of the performance of continuous data. Our study demonstrates the problem surrounding the efficacy of continuous data as remaining relatively intractable despite an exhaustive search, due in part to the difficulty in obtaining relevant comparisons from the literature. Thus, we implore researchers to address this issue with studies that collect discrete and continuous data sets with directly comparable properties (i.e., describing shape, or size).

Evaluation of super resolution technology for digestive endoscopic images

ObjectThis study aims to evaluate the value of super resolution (SR) technology in augmenting the quality of digestive endoscopic images. MethodsIn the retrospective study, we employed two advanced SR models, i.e., SwimIR and ESRGAN. Two discrete datasets were utilized, with training conducted using the dataset of the First Affiliated Hospital of Soochow University (12,212 high-resolution images) and evaluation conducted using the HyperKvasir dataset (2,566 low-resolution images). Furthermore, an assessment of the impact of enhanced low-resolution images was conducted using a 5-point Likert scale from the perspectives of endoscopists. Finally, two endoscopic image classification tasks were employed to evaluate the effect of SR technology on computer vision (CV). ResultsSwinIR demonstrated superior performance, which achieved a PSNR of 32.60, an SSIM of 0.90, and a VIF of 0.47 in test set. 90 % of endoscopists supported that SR preprocessing moderately ameliorated the readability of endoscopic images. For CV, enhanced images bolstered the performance of convolutional neural networks, whether in the classification task of Barrett's esophagus (improved F1-score: 0.04) or Mayo Endoscopy Score (improved F1-score: 0.04). ConclusionsSR technology demonstrates the capacity to produce high-resolution endoscopic images. The approach enhanced clinical readability and CV models’ performance of low-resolution endoscopic images.

A comprehensive review of deep learning-based models for heart disease prediction

Heart disease (HD) is one of the leading causes of death in humans, posing a heavy burden on society, families, and patients. Real-time prediction of HD can reduce mortality rates and is crucial for timely intervention and treatment of HD. Deep learning (DL)-related methods have higher accuracy and real-time performance in predicting HD. In this study, we comprehensively compared and evaluated the contributions and limitations of DL algorithms, extended deep learning (ETDL) algorithms, and integrated deep learning (integrated DL) algorithms that combine DL with other technologies for predicting HD. The articles considered span the period from 2018 to 2023, and after rigorous screening, 64 articles were selected for preliminary research. A systematic literature review of real-time HDP will provide future researchers with a comprehensive understanding of existing deep learning methods and related integrated technologies in the healthcare industry. Furthermore, it discusses popular datasets employed in deploying numerous prediction models. Additionally, it reveals existing open problems or challenges encountered by previous researchers. Notably, the most prevalent challenge is the scarcity of large discrete datasets, followed by the need for further improvement of existing models.

A Novel generalization of sequential decision-theoretic rough set model and its application

This paper introduces a refined and broadened version of decision-theoretic rough sets (DTRSs) named Generalized Sequential Decision-Theoretic Rough Set (GSeq-DTRS), which integrates the three-way decision (3WD) methodology. Operating through multiple levels, this iterative approach aims to comprehensively explore the boundary region. It introduces the concept of generalized granulation, departing from conventional equivalence-relation-based structures to incorporate similarity/tolerance relations. GSeq-DTRS addresses the limitations encountered by its predecessor, Seq-DTRS, particularly in managing sequential procedures and integrating new attributes. A notable advancement is its seamless handling of continuous-scale datasets through a defined Generalized Granular Structure (GGS), enabling classification across multiple levels without necessitating reduction of attributes. A refined version of conditional probability (CP), aligned with tolerance classes, enhances the approach, supported by a meticulously designed algorithm. Extensive experimental analysis conducted on two datasets sourced from https://www.kaggle.com demonstrates the efficacy of the procedure for both continuous and discrete datasets, effectively classifying probable elements into POS or NEG regions based on their membership. Unlike traditional Seq-DTRS, it does not require reduction of attributes at each new level. Additionally, the algorithm exhibits lower sensitivity to parametric values and yields results in fewer iterations. Thus, its potential impact on decision-making processes is readily apparent.

Future Prediction for Tax Complaints to Turkish Ombudsman by Models from Polynomial Regression and Parametric Distribution

The aim of this study is to forecast the amount of tax complaints filed with the Turkish Ombudsman in the future and whether or not policymakers require a specific tax Ombudsman. The polynomial regression for discrete data set is proposed to fit the number of events of tax complaints in the period from years $2013$ to $2021$. The artificial data set is generated by models which are polynomial regression and parametric distribution. The location, scale and shape parameters are determined according to the smallest value between the observed and predicted dependent variable. After determining the smallest value for the tried values of shape parameter and the parameters of polynomial regression, the best value determined by grid search for shape parameter is around $1.07$. Thus, the heavy-tailed from of exponential power distribution is gained. The artificial data sets are generated and sorted from the smallest to biggest ones. The maximum values are around $700$ and $800$ which can be regarded as future prediction because the distance among observations is taken into account by models from polynomial regression and parametric distribution. Since the polynomial regression and the parametric models are used simultaneously for modelling, the distance among observations can also be modelled by parametric model as an alternative approach provided.

An incremental singular value decomposition approach for large-scale spatially parallel & distributed but temporally serial data – applied to technical flows

The article presents a strategy and its algorithm to compile a simulation-accompanying, incremental Singular Value Decomposition (SVD) for time-evolving, spatially parallel discrete data sets. The framework addresses state-of-the-art PDE solvers for computational science and engineering applications. An important characteristic of such applications is that the spatial size of the data is often time-invariant and significantly exceeds the temporal size due to the large computational grid in 3D applications. Typical examples, which are also considered in this article, relate to results extracted from unsteady flow simulations. Herein, the flow data, which progresses over time, is frequently calculated spatially parallel based on domain decomposition strategies, which allow to parallelize the simulation on distributed memory machines following a Single Instruction Multiple Data (SIMD) concept.With a view to the memory-efficient reuse of (compressed) simulation results and their CPU time-saving, sufficiently accurate generation, the paper scrutinizes the efficiency of incremental/parallel SVD approaches for such simulation examples. To improve the computational efficiency, the introduction of a bunch matrix is proposed, which enables the aggregation of multiple time steps and SVD updates, and significantly increases the efficiency. The suggested strategy is verified and validated by simple 2D laminar single-phase flows and subsequently applied to more complex 2D and 3D turbulent two-phase flows. Emphasis is given to (a) the accuracy of SVD-based reconstruction, (b) the physical realizability of the reconstructed quantities, (c) the independence of domain partitioning, (d) an efficient snapshot bunching, and (e) related implementation aspects. In addition, the influence of lower and (adaptive) upper rank thresholds on the effort and accuracy is evaluated. A final application renders the practical benefits of the approach and refers to a merchant ship in head waves at Re = 1.4×107 and Fn = 0.26. The simulation involves 2880 processor cores and the related full-rank snapshot matrix has (108×104) entries. With a numerical overhead of O(10%), this snapshot matrix can be incrementally generated and compressed by O(95%). The compression is accompanied by only small errors in the integral force and local wave elevation of O(10−2%). This qualifies the method for an efficient subsequent data processing.

Spatial assessment of the soil organic carbon content under different types of land use in the Ohrid valley

Spatial assessment of key soil properties is a basic prerequisite for the evidence-based decision making and sustainable use and management of soil. The aim of this work was to estimate the spatial distribution of SOC under different types of land use, by the means of Digital Soil Mapping techniques. A site-specific soil data collection for the Ohrid valley was integrated with continuous and discrete datasets of environmental covariates, serving as predictors. The selected test area outlines the variability of factors influencing the SOC content and spatial distribution. Soil sampling locations were randomly distributed within a predefined mesh with a 1-sq.km spatial resolution and further stratified to outline different types of land use within each mash square. Soil samples were collected from 93 locations at three depths, each 20 cm apart, covering the total area of 10 thousand ha of arable land, forestland, and land under natural vegetation. A set of additional environmental dataset was collected, namely the soil map, land use map, geology map, digital terrain model and its derivatives, satellite images, climate data, as well as a set of indices NDVI, SAVI, BI etc., developed from the remote sensing datasets. Multiple linear regression was used for evaluating the regression pattern between the environmental predictors and the target variable. To estimate spatial variability, several regression tree methods were used. The results obtained using this approach have given a better spatial overview of the most vulnerable areas regarding SOC depletion. Out of 21 locations examined, the content of soil organic carbon in the top layer (0-20 cm.) of forestland was on average 6.81%, while at 22 locations examined under grassland, the average content was 4.07%. The arable land, which is under continuous human impact, had the lowest content of SOC of 2.5% under field crops and 2.61% under perennials.

Identifying Earthquakes in Low-Cost Sensor Signals Contaminated with Vehicular Noise

The importance of monitoring earthquakes for disaster management, public safety, and scientific research can hardly be overstated. The emergence of low-cost seismic sensors offers potential for widespread deployment due to their affordability. Nevertheless, vehicular noise in low-cost seismic sensors presents as a significant challenge in urban environments where such sensors are often deployed. In order to address these challenges, this work proposes the use of an amalgamated deep neural network constituent of a DNN trained on earthquake signals from professional sensory equipment as well as a DNN trained on vehicular signals from low-cost sensors for the purpose of earthquake identification in signals from low-cost sensors contaminated with vehicular noise. To this end, we present low-cost seismic sensory equipment and three discrete datasets that—when the proposed methodology is applied—are shown to significantly outperform a generic stochastic differential model in terms of effectiveness and efficiency.

Lead identification against Mycobacterium tuberculosis using highly enriched active molecules against pantothenate synthetase

The Pantothenate synthetase (PS) from the Mycobacterium tuberculosis (Mtb) holds a crucial role in the survival and robust proliferation of bacteria through its catalysis of coenzyme A and acyl carrier protein synthesis. The present study undertook the PS drug target in complex with a co-crystallized ligand and subjected it to docking and virtual screening approaches. The experimental design encompassed three discrete datasets: an active dataset featuring 136 compounds, an inactive dataset comprising 56 compounds, and a decoys dataset curated from the zinc library, comprising an extensive compilation of approximately 53,000 compounds. The compounds’ binding energies were observed to be in the range of −5 to ∼−14 kcal/mol. Additionally, binding energy results were further refined through Enrichment Factor analysis (EF). EF is a new statistical approach which uses the scores obtained from docking-based virtual screening and predicts the precision of the scoring function. Remarkably, the Enrichment Factor (EF) analysis produced exceptionally favorable outcomes, attaining an EF of approximately 49% within the uppermost 1% fraction of the compound distribution. Finally, a total of eight compounds, evenly distributed between the active dataset and the decoys dataset, emerged as potent inhibitors of the Pantothenate synthetase (PS) enzyme. The analysis of inhibition constants and binding energy revealed a notable correlation, with an r-squared value (r 2) of 0.912 between the two parameters. Furthermore, the shortlisted compounds were subjected to 100 ns MD simulation to determine their stability and dynamics behavior. The decoy compounds that have been identified, exhibiting properties comparable to the active compounds, are postulated as potential candidates for targeting the Pantothenate synthetase (PS) enzyme to treat Mtb infection. Nevertheless, in the pursuit of a comprehensive investigation, it is advisable to undertake additional experimental validation as a component of the subsequent study. Communicated by Ramaswamy H. Sarma.

Crack density estimation method for composite laminates based on discrete form stiffness degradation models

Matrix cracking of composite laminates leads to stiffness degradation and dominates the medium leakage of liner-less composite vessels. A stiffness degradation model can be established through the machine learning method, which provides only discrete form equivalent moduli data without explicit analytical results. This paper further explored the use of the discrete form data set of stiffness degradation in the crack density estimation for composite laminates by using an iterative process. Furthermore, different mixed-mode fracture criteria were discussed. The results showed that Benzeggagh-Kenane (B-K) criterion and the power law criterion are more appropriate in mixed-mode crack density analyses.

TWO-PARTITION DISCRETE GLOBAL GRID SYSTEMS - A COMPARISON OF APPROACHES BASED ON ADJUSTED SPHERICAL CUBE AND CYLINDRICAL EQUIDISTANT PROJECTION

The enormous volumes of geospatial data and the need to process and distribute them cry out for a unified framework that enables their efficient storage, analysis, and a high degree of interoperability. Discrete global grid systems provide such a framework by hierarchically tessellating cells to seamlessly partition and address the globe. Since they are usually based on a regular polyhedron, they partition the entire world into as many discrete data sets as the given polyhedron has sides. In this paper, we try to reduce the number of partitions to two, which is a minimum if we want to obtain spatially convex partitions without interruptions. Two approaches are presented, based on an adjusted spherical cube and an equidistant cylindrical projection. The distortions resulting from the application of these projections are compared and guidelines are presented to improve the quality of their implementation by reducing the distortion of the continental plates and making a better mapping to the WGS84 ellipsoid.

Data‐driven 1D wave propagation for site response analysis

AbstractPrediction of ground motion triggered by earthquakes is a prime concern for both the seismology community and geotechnical earthquake engineering one. The subfield occupied with such a problem is termed site response analysis (SRA), its one‐dimensional flavor (1D‐SRA) being particularly popular given its simplicity. Despite the simple geometrical setting, a paramount challenge remains when it comes to numerically consider intense shaking in the soft upper soil strata of the crust: how to mathematically model the high‐strain, dissipative, potentially rate‐dependent soil behavior. Both heuristics models and phenomenological constitutive laws have been developed to meet the challenge, neither of them being exempt of either numerical limitations or physical inadequacies or both. We propose herein to bring the novel data‐driven paradigm to bear, thus giving away with the need to construct constitutive behavior models altogether. Data‐driven computational mechanics (DDCM) is a novel paradigm in solid mechanics that is gaining popularity; in particular, the multiscale version of it relies on studying the response of the microstructure (in the case of soil, representative volumes containing grains) to populate a dataset that is later used to inform the response at the macroscale. This article presents the first application of multiscale DDCM to 1D‐SRA: first, we demonstrate its capacity to handle wave propagation problems using discrete datasets, obtained via sampling grain ensembles using the discrete element method (DEM), in lieu of a constitutive law and then we apply it specifically to analyze the propagation of harmonic waves in a soft soil deposit that overlies rigid bedrock. The soil in this application displays elastic yet non‐linear response that is depth‐dependent due to evolving overburden pressure. We validate the implementation via comparison to regular finite elements analyses (FEA), discuss the benefits of using DD, and demonstrate that traditional amplification functions are recovered when using the DDCM. Finally, we discuss a number of future work opportunities that lie ahead of this proof‐of‐concept, chiefly in terms of site‐specific studies and incorporating more complex realistic traits of soil behavior. This project was developed using open‐source software and the relevant code is freely made available to other researchers.

Global‐Krigger: A Global Kriging Interpolation Toolbox With Paleoclimatology Examples

AbstractMany applications in Earth sciences require spatial prediction, that is, obtaining a continuous scalar field from a set of discrete scalar data points on the Earth's surface. Such applications include model‐data comparisons and derivation of continuous scalar fields as input for Earth system models. The advantage of kriging as an interpolation method is that it provides predictions with confidence intervals for data sets of irregularly distributed points in space. However, the theory of kriging for non‐Euclidean domains such as oblate spheroids (e.g., the Earth's surface) is poorly developed, and existing kriging algorithms for global interpolation oftentimes cannot guarantee the validity of their predictions. Here, we present Global‐Krigger, a new kriging interpolation algorithm adapted for local to global applications that (a) incorporates a numerical check to guarantee that the necessary conditions for the kriging system of linear equations are met, and (b) derives a combined uncertainty field due both to spatial variations in data density and measurement error. The robustness of the method is demonstrated by cross‐validating predictions against reanalysis fields of traditional climatological scalar variables. We also show an example application in paleoclimatology for Holocene mineral dust deposition fluxes. The toolbox includes a user‐friendly graphical user interface that guides users through a range of choices during data pre‐interpolation analysis, kriging, and post‐processing.

Phase computation for the finite-genus solutions to the focusing nonlinear Schrödinger equation using convolutional neural networks

We develop a method for retrieving a set of parameters of a quasi-periodic finite-genus (finite-gap) solution to the focusing nonlinear Schrödinger (NLS) equation, given the solution evaluated on a finite spatial interval for a fixed time. These parameters (named “phases”) enter the jump matrices in the Riemann-Hilbert (RH) problem representation of finite-genus solutions. First, we detail the existing theory for retrieving the phases for periodic finite-genus solutions. Then, we introduce our method applicable to the quasi-periodic solutions. The method is based on utilizing convolutional neural networks optimized by means of the Bayesian optimization technique to identify the best set of network hyperparameters. To train the neural network, we use the discrete datasets obtained in an inverse manner: for a fixed main spectrum (the endpoints of arcs constituting the contour for the associated RH problem) and a random set of modal phases, we generate the corresponding discretized profile in space via the solution of the RH problem, and these resulting pairs – the phase set and the corresponding discretized solution in a finite interval of space domain – are then employed in training. The method’s functionality is then verified on an independent dataset, demonstrating our method’s satisfactory performance and generalization ability.

Extended Smoothing Methods for Sparse Test Data Based on Zero-Padding

Aiming at the problem of sparse measurement points due to test conditions in engineering, a smoothing method based on zero-padding in the wavenumber domain is proposed to increase data density. Firstly, the principle of data extension and smoothing is introduced. The core idea of this principle is to extend the discrete data series by zero-padding in the wavenumber domain. The conversion between the spatial and wavenumber domains is achieved using the Discrete Fourier Transform (DFT) and the Inverse Discrete Fourier Transform (IDFT). Then, two sets of two-dimensional discrete random data are extended and smoothed, respectively, and the results verify the effectiveness and feasibility of the algorithm. The method can effectively increase the density of test data in engineering tests, achieve smoothing and extend the application to areas related to data processing.

Analytical high-dimensional operators in canonical polyadic finite basis representation (CP-FBR).

In the present work, we introduce a simple means of obtaining an analytical (i.e., grid-free) canonical polyadic (CP) representation of a multidimensional function that is expressed in terms of a set of discrete data. For this, we make use of an initial CP guess, even not fully converged, and a set of auxiliary basis functions [finite basis representation (FBR)]. The resulting CP-FBR expression constitutes the CP counterpart of our previous Tucker sum-of-products-FBR approach. However, as is well-known, CP expressions are much more compact. This has obvious advantages in high-dimensional quantum dynamics. The power of CP-FBR lies in the fact that it requires a grid much coarser than the one needed for the dynamics. In a subsequent step, the basis functions can be interpolated to any desired density of grid points. This is useful, for instance, when different initial conditions (e.g., energy content) of a system are to be considered. We show the application of the method to bound systems of increased dimensionality: H2 (3D), HONO (6D), and CH4 (9D).

Prevalence and predictors of sleep problems in women following a cancer diagnosis: results from the women's wellness after cancer program.

Using a discrete dataset from the Women's Wellness after Cancer Program (WWACP), we examine the prevalence and predictors of self-reported sleep problems in women previously treated for cancer. Participants were 351 women (Mage = 53.2, SD = 8.8) from the WWACP who had completed surgery, chemotherapy and/or radiotherapy for breast, gynaecological or blood cancers within the previous 24months. Sleep problems were measured using the Pittsburgh Sleep Quality Index (PSQI). Baseline data (i.e. prior to intervention randomisation) were analysed. Most women (59%) reported clinically significant sleep disturbance (PSQI > 5), 40% reported insufficient sleep duration (< 7h), 38% self-reported poor sleep quality and 28% reported poor habitual sleep efficiency (sleep efficiency < 75%). Fewer psychological and vasomotor climacteric symptoms, age < 45years and having a partner were associated with reduced odds (AOR < 1) of sleep problems. Higher levels of pain-related disability, and an intermediate compared to 'high' level of education, were associated with increased odds (AOR > 1) of sleep problems. These findings confirm previous studies that have found a high prevalence of sleep problems in women previously treated for cancer. A range of sociodemographic, climacteric and pain-related factors were associated with sleep problemsin this study. Targeted interventions to improve sleep quality after cancer treatment should be explored in this population. Predictors identified in this study could inform intervention targeting and development.

A Fuzzy-Based Fast Feature Selection Using Divide and Conquer Technique in Huge Dimension Dataset

Feature selection is commonly employed for identifying the top n features that significantly contribute to the desired prediction, for example, to find the top 50 or 100 genes responsible for lung or kidney cancer out of 50,000 genes. Thus, it is a huge time- and resource-consuming practice. In this work, we propose a divide-and-conquer technique with fuzzy backward feature elimination (FBFE) that helps to find the important features quickly and accurately. To show the robustness of the proposed method, it is applied to eight different datasets taken from the NCBI database. We compare the proposed method with seven state-of-the-art feature selection methods and find that the proposed method can obtain fast and better classification accuracy. The proposed method will work for qualitative, quantitative, continuous, and discrete datasets. A web service is developed for researchers and academicians to select top n features.

Tracer-field cross-correlations with k-nearest neighbour distributions

ABSTRACT In astronomy and cosmology significant effort is devoted to characterizing and understanding spatial cross-correlations between points – e.g galaxy positions, high energy neutrino arrival directions, X-ray and AGN sources, and continuous fields – e.g. weak lensing meiand Cosmic Microwave Background maps. Recently, we introduced the k-nearest neighbour (kNN) formalism to better characterize the clustering of discrete (point) data sets. Here, we extend it to the point – field cross-correlations analysis. It combines kNN measurements of the point data set with measurements of the field smoothed at many scales. The resulting statistics are sensitive to all orders in the joint clustering of the points and the field. We demonstrate that this approach, unlike the 2-pt cross-correlation, can measure the statistical dependence of two data sets even when there are no linear (Gaussian) correlations between them. We further demonstrate that this framework is far more effective than the two point function in detecting cross-correlations when the continuous field is contaminated by high levels of noise. For a particularly high level of noise, the cross-correlation between haloes and the underlying matter field in a cosmological simulation, between 10 h−1 Mpc and 30 h−1 Mpc, is detected at &amp;gt;5σ significance using the technique presented here, when the two-point cross-correlation significance is ∼1σ. Finally, we show that kNN cross-correlations of haloes and the matter field can be well modelled on quasi-linear scales using the Hybrid Effective Field Theory (HEFT) framework, with the same set of bias parameters as are used for 2-pt cross-correlations. The substantial improvement in the statistical power of detecting cross-correlations using this method makes it a promising tool for various cosmological applications.

Treatment of missing data in Bayesian network structure learning: an application to linked biomedical and social survey data

BackgroundAvailability of linked biomedical and social science data has risen dramatically in past decades, facilitating holistic and systems-based analyses. Among these, Bayesian networks have great potential to tackle complex interdisciplinary problems, because they can easily model inter-relations between variables. They work by encoding conditional independence relationships discovered via advanced inference algorithms. One challenge is dealing with missing data, ubiquitous in survey or biomedical datasets. Missing data is rarely addressed in an advanced way in Bayesian networks; the most common approach is to discard all samples containing missing measurements. This can lead to biased estimates. Here, we examine how Bayesian network structure learning can incorporate missing data.MethodsWe use a simulation approach to compare a commonly used method in frequentist statistics, multiple imputation by chained equations (MICE), with one specific for Bayesian network learning, structural expectation-maximization (SEM). We simulate multiple incomplete categorical (discrete) data sets with different missingness mechanisms, variable numbers, data amount, and missingness proportions. We evaluate performance of MICE and SEM in capturing network structure. We then apply SEM combined with community analysis to a real-world dataset of linked biomedical and social data to investigate associations between socio-demographic factors and multiple chronic conditions in the US elderly population.ResultsWe find that applying either method (MICE or SEM) provides better structure recovery than doing nothing, and SEM in general outperforms MICE. This finding is robust across missingness mechanisms, variable numbers, data amount and missingness proportions. We also find that imputed data from SEM is more accurate than from MICE. Our real-world application recovers known inter-relationships among socio-demographic factors and common multimorbidities. This network analysis also highlights potential areas of investigation, such as links between cancer and cognitive impairment and disconnect between self-assessed memory decline and standard cognitive impairment measurement.ConclusionOur simulation results suggest taking advantage of the additional information provided by network structure during SEM improves the performance of Bayesian networks; this might be especially useful for social science and other interdisciplinary analyses. Our case study show that comorbidities of different diseases interact with each other and are closely associated with socio-demographic factors.