Wessel N. van Wieringen’s research while affiliated with Vrije Universiteit Amsterdam and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (146)


Figure 1: Set-up of FusedTree. In each leaf node m (m = 1, . . . , 4 in this example), we fit a linear regression using n m samples with omics covariates X (m) and an intercept c m . The intercept contains the (potentially nonlinear) clinical information. The regression in leaf node m borrows information from the other leaf nodes by linking the regressions (indicated with ←→) through fusion penalty (5).
Figure 2: Boxplots of the prediction mean square errors of several prediction models across 500 simulated data sets for the Interaction(top), Full Fusion (middle), and Linear (bottom) simulation experiment. For all experiments, we consider N = 100 (left) and N = 300 (right). The oracle prediction model is only considered for the Interaction experiment ( * indicates that oracle model boxplots are missing for the Full Fusion and Linear experiment). We do not depict results for ridge regression in the Interaction experiment because its PMSE's fall far outside the range of the PMSE's of the other models (indicated by ↑). Outliers of boxplots are not shown.
Figure 3: (a) The estimated survival tree of FusedTree. In the leaf nodes, the relative death rate (top) and the number of events/node sample size (bottom) are depicted. The plot is produced using the R package rpart.plot. (b) Regularization paths as a function of fusion penalty α for the effect estimates of two genes in nodes 5, 12, and 13 of FusedTree. The vertical dotted line (at log α = 9.6) indicates the tuned α of FusedTree.
Figure S1: Fit of the tree
Figure S4: Scatter plot of PMSE ZeroFus /PMSE FusedTree as a function of fusion penalty α (log scale) across 500 simulated data sets for N = 100 and N = 300 for the effect modification simulation experiment (Section 4.1)

+3

Fusion of Tree-induced Regressions for Clinico-genomic Data
  • Preprint
  • File available

November 2024

·

39 Reads

Jeroen M. Goedhart

·

Mark A. van de Wiel

·

Wessel N. van Wieringen

·

Thomas Klausch

Cancer prognosis is often based on a set of omics covariates and a set of established clinical covariates such as age and tumor stage. Combining these two sets poses challenges. First, dimension difference: clinical covariates should be favored because they are low-dimensional and usually have stronger prognostic ability than high-dimensional omics covariates. Second, interactions: genetic profiles and their prognostic effects may vary across patient subpopulations. Last, redundancy: a (set of) gene(s) may encode similar prognostic information as a clinical covariate. To address these challenges, we combine regression trees, employing clinical covariates only, with a fusion-like penalized regression framework in the leaf nodes for the omics covariates. The fusion penalty controls the variability in genetic profiles across subpopulations. We prove that the shrinkage limit of the proposed method equals a benchmark model: a ridge regression with penalized omics covariates and unpenalized clinical covariates. Furthermore, the proposed method allows researchers to evaluate, for different subpopulations, whether the overall omics effect enhances prognosis compared to only employing clinical covariates. In an application to colorectal cancer prognosis based on established clinical covariates and 20,000+ gene expressions, we illustrate the features of our method.

Download

The gray shaded area demarcates the combinations of ωd\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega _d$$\end{document} and ωo\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega _o$$\end{document} that produce a positive definite precision matrix, i.e. the set of feasible estimates. The green dashed and red dashed-dotted lines are the ℓ2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ell _2$$\end{document}-penalized precision and covariance parameter constraint, respectively. Estimates to the ‘right’ of these constraints are excluded (color figure online)
Illustration of the ridge prior on covariance and precision elements. Left panel: p=1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p=1$$\end{document}, σ2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma ^2$$\end{document} has a truncated normal prior (tσ=2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t_{\sigma }=2$$\end{document} and λσ=10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda _{\sigma } = 10$$\end{document}) and the implied prior on its inverse σ-2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma ^{-2}$$\end{document} is displayed (black line). Right panel: The toy example of Sect. 2 is considered where Ω=(ωd-ω0)I33+ω0133\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varvec{\Omega } = (\omega _d - \omega _0) {\textbf {I}}_{33} + \omega _{0} {\textbf {1}}_{33}$$\end{document} is parameterized in terms of two parameters only. The contour plot of the ridge prior where Tω=033\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\textbf {T}}_\omega = \varvec{0}_{33} $$\end{document} and λω=1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda _{\omega } = 1$$\end{document} is displayed along with the positive definite domain of Ω\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varvec{\Omega }$$\end{document}
Ridge-type covariance and precision matrix estimators of the multivariate normal distribution

October 2024

·

14 Reads

Statistical Papers

We consider ridge-type estimation of the multivariate normal distribution’s covariance matrix and its inverse, the precision matrix. While several ridge-type covariance and precision matrix estimators have been presented in the literature, their respective inverses are often not considered as precision and covariance matrix estimators even though their estimands are one-to-one related through the matrix inverse. We study which estimator is to be preferred in what case. Hereto we compare the ridge-type covariance matrix estimators and their properties to that of the inverse of the ridge-type precision matrix estimators, and vice versa. The comparison, in which we take all ridge-type estimators along, is limited to a specific case that is illustrative of the difference between the covariance and precision matrix estimators. The comparison addresses the estimators’ estimating equation, analytic expression, analytic properties like positive definiteness and penalization limit, mean squared error, consistency, Bayesian formulation, and their loss and potential for marginal and partial correlation screening.


Higher‐order functional connectivity analysis of resting‐state functional magnetic resonance imaging data using multivariate cumulants

March 2024

·

57 Reads

·

6 Citations

·

·

·

[...]

·

Prejaas K. B. Tewarie

Blood‐level oxygenation‐dependent (BOLD) functional magnetic resonance imaging (fMRI) is the most common modality to study functional connectivity in the human brain. Most research to date has focused on connectivity between pairs of brain regions. However, attention has recently turned towards connectivity involving more than two regions, that is, higher‐order connectivity. It is not yet clear how higher‐order connectivity can best be quantified. The measures that are currently in use cannot distinguish between pairwise (i.e., second‐order) and higher‐order connectivity. We show that genuine higher‐order connectivity can be quantified by using multivariate cumulants. We explore the use of multivariate cumulants for quantifying higher‐order connectivity and the performance of block bootstrapping for statistical inference. In particular, we formulate a generative model for fMRI signals exhibiting higher‐order connectivity and use it to assess bias, standard errors, and detection probabilities. Application to resting‐state fMRI data from the Human Connectome Project demonstrates that spontaneous fMRI signals are organized into higher‐order networks that are distinct from second‐order resting‐state networks. Application to a clinical cohort of patients with multiple sclerosis further demonstrates that cumulants can be used to classify disease groups and explain behavioral variability. Hence, we present a novel framework to reliably estimate genuine higher‐order connectivity in fMRI data which can be used for constructing hyperedges, and finally, which can readily be applied to fMRI data from populations with neuropsychiatric disease or cognitive neuroscientific experiments.


Figure 2 -Flowchart of the preprocessing steps of the acceleration signals in the time (white swim lane) and frequency (gray swim line) domains. Note.
Configurations and Number of Accelerometer Recordings Used in Analyses per Experiment
Inter-Brand, -Dynamic Range, and -Sampling Rate Comparability of Raw Accelerometer Data as Used in Physical Behavior Research

January 2024

·

45 Reads

·

2 Citations

Journal for the Measurement of Physical Behaviour

Objective : Previous studies that looked at comparability of accelerometer data focused on epoch or recording level comparability. Our study aims to provide insight into the comparability at raw data level. Methods : We performed five experiments with accelerometers attached to a mechanical shaker machine applying movement along a single axis in the horizontal plane. In each experiment, a 1-min no-movement condition was followed by nineteen 2-min shaker frequency conditions (30–250 rpm). We analyzed accelerometer data from Axivity, ActiGraph, GENEActiv, MOX, and activPAL devices. Comparability between commonly used brands and dynamic ranges was assessed in the frequency domain with power spectra and in the time domain with maximum lagged cross-correlation analyses. The influence of sampling rate on magnitude of acceleration across brands was explored visually. All data were published open access. Results : Magnitude of noise in rest was highest in MOX and lowest in ActiGraph. The signal mean power spectral density was equal between brands at low shaker frequency conditions (<3.13 Hz) and between dynamic ranges within the Axivity brand at all shaker frequency conditions. In contrast, the cross-correlation coefficients between time series across brands and dynamic ranges were higher at higher shaking frequencies. Sampling rate affected the magnitude of acceleration most in Axivity and least in GENEActiv. Conclusions : The comparability of raw acceleration signals between brands and/or sampling rates depends on the type of movement. These findings aid a more fundamental understanding and anticipation of differences in behavior estimates between different implementations of raw accelerometry.



Figure 4: The top row shows the time series chain graph of the toy model. The bottom row displays the three paths in the time series chain graph connecting the first variate between to consecutive time points.
Figure 22: Box plot of specificity (false positive rate) and sensitivity (true positive rate) of the ridge and SCAD methods on simulated data where p=25, T=10, n=5 and A with roughly 5% nonzero elements.
Supplementary Material to: Ridge estimation of the VAR(1) model and its time series chain graph from multivariate time-course omics data

August 2023

·

25 Reads

Biometrical Journal

Omics experiments endowed with a time-course design may enable us to uncover the dynamic interplay among genes of cellular processes. Multivariate techniques (like VAR(1) models describing the temporal and contemporaneous relations among variates) that may facilitate this goal are hampered by the high dimensionality of the resulting data. This is resolved by the presented ridge regularized maximum likelihood estimation procedure for the VAR(1) model. Information on the absence of temporal and contemporaneous relations may be incorporated into this procedure. Its computationally efficient implementation is discussed. The estimation procedure is accompanied by a LOOCV scheme to determine the associated penalty parameters. Downstream exploitation of the estimated VAR(1) model is outlined: an empirical Bayes procedure to identify the interesting temporal and contemporaneous relationships, impulse response analysis, mutual information analysis, and covariance decomposition into the (graphical) relations among variates. In a simulation study, the presented ridge estimation procedure outperformed a sparse competitor in terms of Frobenius loss of the estimates, while their selection properties are on par. The proposed machinery is illustrated in the reconstruction of the p53 signaling pathway during HPV-induced cellular transformation. The methodology is implemented in the ragt2ridges R-package available from CRAN.


Figure 1. Illustration of fusion along the axes (left panel) and along the diagonals (right panel). The dot ( †) represents the array coordinates of the estimate that is shrunken towards to estimates that are at the end of the arrow heads. For both the axes and diagonal fusion, the illustration contains the shrinkage of an 'interior', 'boundary', and 'corner' estimate.
Figure 2. From left to right: the true regression parameter, the regular ridge, and two-dimensional fused ridge regression estimate with optimal cross-validated penalty parameters.
Figure 3. Upper panels: heatmap of the fused ridge (left) and aggregated ridge (right) regression estimates in two-dimensional layout. The labels indicate the values at the grid points, with 'A' and 'L' standing for 'acceleration' and 'length'. For instance, 'L2.18' refers to bouts of exp (2.18) minutes, while 'A6.57' to an acceleration of sinh(6.57) counts per minute. Left and right lower panels: Violin plots of Allen's PRESS statistic and Spearman's rank correlation of the fit and of the prediction with and without the accelerometer data. The 'U only'-label of x-axis of the violin plots corresponds to the model without accelerometer data, i.e. including only classical covariates. The other labels refer to models that include accelerometer data but are fitted with the standard ridge penalty ('ridge'), and the fused ridge one ('fused, 30 dof' and 'fused, 100 dof'). The number in the latter two labels refers to the bound on the maximum degrees of freedom that the estimator could consume.
Two-dimensional fused targeted ridge regression for health indicator prediction from accelerometer data

June 2023

·

52 Reads

·

3 Citations

Journal of the Royal Statistical Society Series C Applied Statistics

We present the two-dimensional targeted fused ridge estimator of the linear and logistic regression models. The estimator (i) handles both unpenalised and penalised covariates, (ii) accommodates possible relations among the covariates’ coefficients through a fusion penalty, and (iii) incorporates prior information on the regression parameter through a non-zero shrinkage target. In this work, the aforementioned relations are similarities among the covariates’ coefficients due to spatial proximity in a two-dimensional grid. In an extensive re-analysis of an epidemiological and an image analysis study, we illustrate the use of the estimator’s aforementioned features that result in a tangibly interpretable predictor.


Network. Thickness of line indicates the median of the strength of the partial correlation. Abbreviations: AD. = use of antidepressants, ADL = activities of daily living, AES = apathy evaluation scale, AP. = use of antipsychotic drugs, BZ. = use of benzodiazepines, CPS = cognitive performance scale, Edu = level of education, IADL = instrumental activities of daily living, Los = length of stay in LCTF, Mansa = Manchester short assessment of quality of life, NPI = neuropsychiatric inventory, NPI-Ag = NPI agitation subscale, NPI-Ap =NPI apathy subscale, NPI-Anx = NPI anxiety subscale, NPI-Psy = NPI psychosis subscale, NPI-Dep = NPI depression subscale, PCRS = patient competency rating scale, RISE = revised index for social functioning.
Network summary statistics: Centrality scores and number of connections.
Cont.
Impaired Awareness in People with Severe Alcohol-Related Cognitive Deficits Including Korskoff’s Syndrome: A Network Analysis

April 2023

·

43 Reads

·

2 Citations

Background: Impaired awareness of one's own functioning is highly common in people with Korsakoff's syndrome (KS). However, it is currently unclear how awareness relates to impairments in daily functioning and quality of life (QoL). Methods: We assessed how impaired awareness relates to cognitive, behavioral, physical, and social functioning and QoL by applying a network analysis. We used cross-sectional data from 215 patients with KS or other severe alcohol-related cognitive deficits living in Dutch long-term care facilities (LTCFs). Results: Apathy has the most central position in the network. Higher apathy scores relate positively to reduced cognition and to a greater decline in activities of daily living and negatively to social participation and the use of antipsychotic drugs. Impaired awareness is also a central node. It is positively related to a higher perceived QoL, reduced cognition and apathy, and negatively to social participation and length of stay in the LTCF. Mediated through apathy and social participation, impaired awareness is indirectly related to other neuropsychiatric symptoms. Conclusions: Impaired awareness is closely related to other domains of daily functioning and QoL of people with KS or other severe alcohol-related cognitive deficits living in LTCFs. Apathy plays a central role. Network analysis offers interesting insights to evaluate the interconnection of different symptoms and impairments in brain disorders such as KS.


rags2ridges : A One-Stop- ℓ 2 -Shop for Graphical Modeling of High-Dimensional Precision Matrices

May 2022

·

18 Reads

·

11 Citations

Journal of Statistical Software

A graphical model is an undirected network representing the conditional independence properties between random variables. Graphical modeling has become part and parcel of systems or network approaches to multivariate data, in particular when the variable dimension exceeds the observation dimension. rags2ridges is an R package for graphical modeling of high-dimensional precision matrices through ridge (ℓ2) penalties. It provides a modular framework for the extraction, visualization, and analysis of Gaussian graphical models from high-dimensional data. Moreover, it can handle the incorporation of prior information as well as multiple heterogeneous data classes. As such, it provides a one-stop-ℓ2-shop for graphical modeling of high-dimensional precision matrices. The functionality of the package is illustrated with an example dataset pertaining to blood-based metabolite measurements in persons suffering from Alzheimer's disease.


Figure 2. The panels show the trajectories of the ML (left) and the updated ridge regression, with its penalty parameter chosen via constrained LOOCV, estimates. Each trajectory represents a single covariate. The presence of a health indicator in the data of a particular year is evident from a symbol on its trajectory at the corresponding year. The symbol is omitted in years that the health indicator was not registered.
Sequential Learning of Regression Models by Penalized Estimation

January 2022

·

23 Reads

·

7 Citations

When data arrive in a sequence of two or more datasets, modeling on the most recent dataset should take previous datasets into account. We specifically investigate a strategy for regression modeling when parameter estimates from previous data can be used as anchoring points, yet may not be available for all parameters, thus, covariance information cannot be reused. A procedure that updates through targeted penalized estimation, which shrinks the estimator toward a nonzero value, is presented. The parameter estimate from the previous data serves as this nonzero value when an update is sought from novel data. This naturally extends to a sequence of datasets with the same response, but potentially only partial overlap in covariates. The iteratively updated regression parameter estimator is shown to be asymptotically unbiased and consistent. The penalty parameter is chosen through constrained cross-validated log-likelihood optimization. The constraint bounds the amount of shrinkage of the updated estimator toward the current one from below. The bound aims to preserve the (updated) estimator’s goodness of fit on all-but-the-novel data. The proposed approach is compared to other regression modeling procedures. Finally, it is illustrated on an epidemiological study where the data arrive in batches with different covariate-availability and the model is refitted with the availability of a novel batch. Supplementary materials for this article are available online.


Citations (64)


... A three-component framework intended to provide a foundational evaluation framework for DHTs has been proposed [17], including three steps: verification, analytical validation, and clinical validation (V3). The engineering verification of our selected sensor system has already been well-described [18]. Here, we used a pilot study to address the analytical validity of potential new digital measures by comparing processed sensor-level data to clinician observation (our criterion measure), and then we explored the application of this approach in a free-living office environment. ...

Reference:

Digital Health Technologies for Optimising Treatment and Rehabilitation Following Surgery: Device-Based Measurement of Sling Posture and Adherence
Inter-Brand, -Dynamic Range, and -Sampling Rate Comparability of Raw Accelerometer Data as Used in Physical Behavior Research

Journal for the Measurement of Physical Behaviour

... The whole-brain models have been used on perturbations and psychiatric or neurological conditions, enabling to test mechanisms that can be used for predicting the outcomes of real experimental settings [37][38][39][40]56]. Higher-order interactions have gained prominence in clinical applications for characterizing and predicting healthy aging [39,57,58], early development [59], neurological conditions [60,61], and their associations with cognition [42] and consciousness [62]. Recently, HOI has been applied to transcranial ultrasound stimulation (TUS) in macaques, revealing different topological reorganizations depending on the stimulation target [25]. ...

Higher‐order functional connectivity analysis of resting‐state functional magnetic resonance imaging data using multivariate cumulants
  • Citing Article
  • March 2024

... To the best of our knowledge, we are the first to use this approach to gain insight into the factors associated with these outcomes. Similar approaches have been used with outcomes such as chronic back pain (Huie et al., 2022) or alcohol-related cognitive deficits (Fidder et al., 2023). As expected, the outcomes severe fatigue, difficulty concentrating, depressive symptoms and limitations in physical functioning, clustered together in the structural network, i.e., were strongly related with each other. ...

Impaired Awareness in People with Severe Alcohol-Related Cognitive Deficits Including Korskoff’s Syndrome: A Network Analysis

... As indicated earlier, Graphical Ridge approaches do not provide sparse estimators. Therefore, we perform an additional sparsification on these estimates using sparsify() function of rags2ridges package, with 'localFDR' thresholding argument (see Peeters et al., 2022, for more details). ...

rags2ridges : A One-Stop- ℓ 2 -Shop for Graphical Modeling of High-Dimensional Precision Matrices

Journal of Statistical Software

... QR decomposition methods that efficiently modify rows in existing datasets prove particularly beneficial in various tasks, including hyper-parameter tuning through cross-validation [64], for evaluating model goodness-of-fit on new data [47], in sequential parameter learning [73,75,104], and to facilitate parallel computing [1,65,62]. These techniques offer computational advantages, especially in high-dimensional problems. ...

Sequential Learning of Regression Models by Penalized Estimation

... In this regard, to explore more and new possible scenarios, it is fundamental and necessary to reduce the computational execution time of the involved simulation processes. Many areas pursue this goal and employ parallel programming for it, for instance in [2][3][4][5][6][7]. ...

A parallel algorithm for ridge-penalized estimation of the multivariate exponential family from data of mixed types

Statistics and Computing

... Before joining them into a common database the data are screened separately for each area of origin with standard statistical methods such as A c c e p t e d M a n u s c r i p t multivariate regression of single outcomes. More suited to unravel the complexity of causal relationships between different levels is bioinformatical network analysis (Schäfer and Strimmer 2005;van Wieringen and Chen 2021). This analytical tool should identify radiation-induced clusters of biomarkers linking molecular and vascular levels. ...

Penalized estimation of the Gaussian graphical model from data with replicates

Statistics in Medicine

... Hussain and Cambria [31] analyzed knowledge-based reasoning using a vector space and support vector machine model, which utilizes lexical and graph representations for sentiment analysis. Aflakparast et al. [150] proposed a Bayesian graphical model to examine Twitter data. Demotte et al. [56] presented a Capsule network-based model that utilizes GloVe embeddings and dynamic/static routing to analyze social media content. ...

Analysis of Twitter data with the Bayesian fused graphical lasso

... Deficiency in Serpin G1/C1-inhibitor also permits kallikrein activation, and the production of the vasoactive peptide bradykinin [50]. In the wound bed, administration of a Serpin G1/C1-inhibitor reduces local inflammation and capillary leakage, although the acceleration of wound closure may depend on the duration of Serpin G1/C1-inhibitor administration [51,52]. With Serpins often classified as acute phase proteins that regulate inflammation, expressing specific Serpins in EVs may promote wound closure by delivering enzyme inhibitors in EVs that have an EVdependent tropism for specific microenvironments of the wound bed. ...

C1 Inhibitor Administration Reduces Local Inflammation and Capillary Leakage, Without Affecting Long-term Wound Healing Parameters, in a Pig Burn Wound Model
  • Citing Article
  • July 2020

Anti-Inflammatory & Anti-Allergy Agents in Medicinal Chemistry