Dynamic multivariate patterns of brain structure-neuropsychiatric symptom associations in long COVID.

Code ↔ Paper

The paper beside its authors' code: matches between them have not been computed for this paper yet.

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

MATLAB · 520 lines · 17 KB · GPL-3.0

function cfg = cfg_defaults(cfg, varargin)
% cfg_defaults
%
% Set defaults in your configuration (`cfg`) structure which will define
% the settings of your analysis (e.g., machine, framework, statistical
% inference). Use this function to update and add all necessary defaults to
% your `cfg`. If you defined anything in your `cfg` before calling the
% function, it won't overwrite those values. The path to the project folder
% should be always defined in your `cfg` or passed as varargin, otherwise
% the function throws an error. All the other fields are optional and can
% be filled up by `cfg_defaults`.
%
% No results will be stored in the cfg structure. See [res_defaults](../res_defaults)
% for more information on results.
%
% !!! note "Warning"
% We strongly advise to inspect the output of `cfg_defaults` to make
% sure that the defaults are set as expected.
%
% # Syntax
% cfg = cfg_defaults(cfg, varargin)
%
% # Inputs
% cfg:: struct
% varargin:: name-value pairs
% additional parameters can be set via name-value pairs with dot notation
% supported (e.g., 'frwork.split.nout', 5)
%
% # Outputs
% cfg:: struct
% configuration structure that has been updated with defaults
%
% # Examples
%
% % Example 1
% [X, Y, wX, wY] = generate_data(1000, 100, 100, 10, 10, 1);
%
% ---
% See also: [cfg](../../cfg), [res_defaults](../res_defaults/)
%
%_______________________________________________________________________
% Copyright (C) 2022 University College London
% Written by Agoston Mihalik ([email hidden])
% $Id$
% This file is part of CCA/PLS Toolkit.
%
% CCA/PLS Toolkit is free software: you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation, either version 3 of the License, or
% (at your option) any later version.
%
% CCA/PLS Toolkit is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details.
%
% You should have received a copy of the GNU General Public License
% along with CCA/PLS Toolkit. If not, see <https://www.gnu.org/licenses/>.
def = parse_input([], varargin{:});
% Initialize cfg
if isempty(cfg)
cfg = struct();
end
cfg = assign_defaults(cfg, def);
%----- Primary defaults
% Get toolkit version from git commit
[~, def.env.commit] = system('git rev-parse HEAD');
% Framework
def.frwork.name = 'holdout'; % holdout/permutation
def.frwork.flag = ''; % string to specify analysis in folder name
% Machine
def.machine.name = 'spls';
% Computer environment
def.env.comp = 'local'; % local/cluster
if ismac
def.env.OS = 'mac';
elseif isunix
def.env.OS = 'unix';
elseif ispc
def.env.OS = 'pc';
end
% Verbosity level
% 1: detailed progress update with elapsed time info
% 2: detailed progress update
% 3: minimal progress update
def.env.verbose = 2;
% Data block structure (see exchangeability blocks below)
def.data.block = 0;
% Data modalities and confounds
def.data.mod = {'X' 'Y'};
def.data.conf = 0;
% Update fields
cfg = assign_defaults(cfg, def);
% Check that path to project folder exists
if ~isfield(cfg, 'dir') || ~isfield(cfg.dir, 'project')
error('Path to project folder should be given.')
end
%----- Secondary defaults
% Data train-test splitting
switch cfg.frwork.name
case 'holdout'
% Multiple holdout framework (see Monteiro et al 2016 J Neurosci Methods)
def.frwork.split.nout = 5;
def.frwork.split.propout = 0.2;
def.frwork.split.nin = 5;
def.frwork.split.propin = 0.2;
case 'permutation'
% Permutation framework (see Smith et al Nat Neurosci 2015)
% without train-test splitting
def.frwork.split.nout = 1;
end
% Number of permutations in statistical inference
def.stat.nperm = 1000;
% Metrics to evaluate machines
def.machine.metric = {'trcorrel' 'correl'}; % correlation between projections
if strcmp(cfg.machine.name, 'spls')
def.machine.metric = [def.machine.metric {'simwx' 'simwy'}]; % stability of weights
end
% Update fields
cfg = assign_defaults(cfg, def);
% Similarity metric
if any(contains(cfg.machine.metric, 'sim'))
if strcmp(cfg.machine.name, 'spls')
def.machine.simw = 'overlap-corrected';
else
def.machine.simw = 'correlation-Pearson';
end
end
% Statistical inference
if cfg.stat.nperm ~= 0
def.stat.crit = 'correl';
def.stat.perm = 'train+test'; % 'train'
def.stat.alpha = 0.05;
end
% Update fields
cfg = assign_defaults(cfg, def);
% Deflation formulation
if ~isfield(cfg.frwork, 'nlevel') || cfg.frwork.nlevel > 1
if ismember(cfg.machine.name, {'pls' 'spls'})
def.defl.name = 'pls-modeA';
else
cfg.defl.name = 'generalized'; % overwrite user's option to be safe
end
% Deflation strategy with data splitting
def.defl.crit = 'correl';
end
% Filename suffix on cluster or set default
if strcmp(cfg.env.comp, 'cluster') && ~isempty(getenv('SGE_TASK_ID')) ...
&& ~strcmp(getenv('SGE_TASK_ID'), 'undefined')
def.env.fileend = ['_' num2str(getenv('SGE_TASK_ID'))];
elseif strcmp(cfg.env.comp, 'cluster') && ~isempty(getenv('SLURM_ARRAY_TASK_ID'))
def.env.fileend = ['_' num2str(getenv('SLURM_ARRAY_TASK_ID'))];
else
def.env.fileend = '_1';
end
% Compression setting for saving files
def.env.save.compression = 1;
% Exchangeability blocks (EB) for restricted partitioning and permutation
% (see Winkler et al 2015 Neuroimage, https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/PALM)
if cfg.data.block
load(fullfile(cfg.dir.project, 'data', 'EB.mat'))
if ~exist('EB', 'var')
error('EB matrix not available');
end
% Columns of EB for restricted partitioning
def.data.EB.split = 1:size(EB, 2);
% Columns of EB for restricted permutations
def.data.EB.stat = 1:size(EB, 2);
end
% Data path and preprocessing
if cfg.data.conf
mod = [cfg.data.mod {'C'}]; % modalities including confound matrix
else
mod = cfg.data.mod;
end
for i=1:numel(mod)
def.data.(mod{i}).fname = fullfile(cfg.dir.project, 'data', [mod{i} '.mat']);
def.data.(mod{i}).preproc = {'impute' 'zscore'};
end
if cfg.data.conf
for i=1:numel(cfg.data.mod)
def.data.(cfg.data.mod{i}).preproc = [def.data.(cfg.data.mod{i}).preproc {'deconf'}];
end
end
% Update fields
cfg = assign_defaults(cfg, def);
% Data processing in details
for i=1:numel(mod)
if ismember('impute', cfg.data.(mod{i}).preproc)
def.data.(mod{i}).impute = 'median';
end
if ismember('deconf', cfg.data.(mod{i}).preproc)
def.data.(mod{i}).deconf = 'standard';
end
end
% Seed options for reproducibility
% this is important in case multiple jobs are started simultaneously on a
% cluster, avoids jobs running same hyperparameters and permutations
% simply uses the number of the fileend as seed
% default, shuffle, number or struct returned by rng
def.env.seed.split = 'default';
tmp = regexp(cfg.env.fileend, '_(\d+)\>', 'tokens');
def.env.seed.model = cellfun(@str2num, tmp{1});
def.env.seed.perm = 'default';
% Load data
if exist(cfg.data.X.fname, 'file') && exist(cfg.data.Y.fname, 'file')
for m=1:numel(mod)
% Access data without loading into memory
matobj = matfile(cfg.data.(mod{m}).fname);
if ismember(who(matobj), mod{m})
data.(mod{m}) = matobj;
else
error('%s matrix not available in %s.', mod{m}, cfg.data.(mod{m}).fname);
end
end
else
error('X and Y input data cannot be found in path, check your data folder.');
end
% Update fields
cfg = assign_defaults(cfg, def);
%----- Defaults set based on X, Y
% Maximum number of associative effects
if strcmp(cfg.defl.name, 'pls-regression')
def.frwork.nlevel = min(size(data.X, 'X'));
else
def.frwork.nlevel = min(cellfun(@(x) min(size(data.(x), x)), cfg.data.mod));
end
if isfield(cfg.frwork, 'nlevel') && cfg.frwork.nlevel > def.frwork.nlevel
cfg.frwork.nlevel = def.frwork.nlevel;
end
% Data dimensionality
for i=1:numel(cfg.data.mod)
if ~isfield(cfg.data, 'nsubj')
cfg.data.nsubj = size(data.(cfg.data.mod{i}), cfg.data.mod{i}, 1);
elseif size(data.(cfg.data.mod{i}), cfg.data.mod{i}, 1) ~= cfg.data.nsubj
error('The number of examples do not match across data modalities');
end
cfg.data.(cfg.data.mod{i}).nfeat = size(data.(cfg.data.mod{i}), cfg.data.mod{i}, 2);
end
% Initialize machine hyperparameters
def.machine.param.name = {};
% Update fields
cfg = assign_defaults(cfg, def);
% Grid search settings for machines
for i=1:numel(cfg.data.mod)
m = lower(cfg.data.mod{i}); % shorthand for modality in lowercase
switch cfg.machine.name
case {'pls' 'spls'}
% Hyperparameter name
if isempty(cfg.machine.param.name)
cfg.machine.param.name = {['L1' m]};
elseif ~ismember(['L1' m], cfg.machine.param.name)
cfg.machine.param.name = [cfg.machine.param.name {['L1' m]}];
end
if strcmp(cfg.machine.name, 'pls')
% Hyperparameter type
cfg.machine.param.type = 'matched';
% No grid search as L1 regularizaion used with fix
% parameter outside of SPLS range
elseif strcmp(cfg.machine.name, 'spls')
% Hyperparameter type
def.machine.param.type = 'factorial';
% L1 regularization hyperparameter
def.machine.param.(['rangeL1' m]) = [1 sqrt(size(data.(upper(m)), upper(m), 2))];
end
case 'cca'
% Hyperparameter name and type
if isempty(cfg.machine.param.name)
cfg.machine.param.name = {['L2' m]};
elseif ~ismember(['L2' m], cfg.machine.param.name)
cfg.machine.param.name = [cfg.machine.param.name {['L2' m]}];
end
def.machine.param.type = 'factorial';
% Explained variance treated as hyperparameter
if ismember(['VAR' m], cfg.machine.param.name)
def.machine.param.(['rangeVAR' m]) = [0.1 1];
end
% PCA components treated as hyperparameter
if ismember(['PCA' m], cfg.machine.param.name)
dim = size(data.(upper(m)), upper(m));
if strcmp(cfg.frwork.name, 'holdout')
dim(1) = dim(1) * (1 - cfg.frwork.split.propout) * (1 - cfg.frwork.split.propin);
end
def.machine.param.(['rangePCA' m]) = [1 floor(min(dim))];
end
case 'rcca'
% Hyperparameter name and type
if isempty(cfg.machine.param.name)
cfg.machine.param.name = {['L2' m]};
elseif ~ismember(['L2' m], cfg.machine.param.name)
cfg.machine.param.name = [cfg.machine.param.name {['L2' m]}];
end
def.machine.param.type = 'factorial';
% L2 regularization hyperparameter
if strcmp(cfg.machine.name, 'rcca')
def.machine.param.(['rangeL2' m]) = [1 size(data.(upper(m)), upper(m), 2)^2];
end
end
end
% Update fields
cfg = assign_defaults(cfg, def);
% Number of default hyperparameters
for i=1:numel(cfg.machine.param.name)
if isfield(cfg.machine.param, (['range' cfg.machine.param.name{i}]))
def.machine.param.(['n' cfg.machine.param.name{i}]) = 10;
end
end
% Update fields
cfg = assign_defaults(cfg, def);
% Hyperparameter settings for machines
for i=1:numel(cfg.data.mod)
m = lower(cfg.data.mod{i}); % shorthand for modality in lowercase
switch cfg.machine.name
case 'pls'
% Uses SPLS with L1 regularization outside of range (see below)
cfg.machine.param.(['L1' m]) = size(data.(upper(m)), upper(m), 2);
case 'spls'
% L1 regularization hyperparameter for sparsity
% Note: logarithmic space for efficient L1 norm
if ismember(['L1' m], cfg.machine.param.name)
def.machine.param.(['L1' m]) = logspace(log10(cfg.machine.param.(['rangeL1' m])(1)), ...
log10(cfg.machine.param.(['rangeL1' m])(2)), cfg.machine.param.(['nL1' m]));
end
case 'cca'
% RCCA with L2 regularization set to 0 (see above)
cfg.machine.param.(['L2' m]) = 0;
% Explained variance treated as hyperparameter
if ismember(['VAR' m], cfg.machine.param.name)
def.machine.param.(['VAR' m]) = linspace(cfg.machine.param.(['rangeVAR' m])(1), ...
cfg.machine.param.(['rangeVAR' m])(end), cfg.machine.param.(['nVAR' m]));
end
% PCA components treated as hyperparameter
if ismember(['PCA' m], cfg.machine.param.name)
def.machine.param.(['PCA' m]) = logspace(log10(cfg.machine.param.(['rangePCA' m])(1)), ...
log10(cfg.machine.param.(['rangePCA' m])(2)), cfg.machine.param.(['nPCA' m]));
def.machine.param.(['PCA' m]) = unique(round(def.machine.param.(['PCA' m])));
end
case 'rcca'
% L2 regularization hyperparameter for smoothing between CCA (l2=0) and PLS (l2=1)
% Note: 1 - logarithmic space for efficient L2 norm
def.machine.param.(['L2' m]) = 1 - logspace(-log10(cfg.machine.param.(['rangeL2' m])(1)), ...
-log10(cfg.machine.param.(['rangeL2' m])(2)), cfg.machine.param.(['nL2' m]));
end
end
% Criterion for hyperparameter selection
def.machine.param.crit = 'correl';
% Additional settings for machines
if ismember(cfg.machine.name, {'pls' 'spls'})
% Tolerance and maximum number of iterations
def.machine.spls.tol = 1e-5;
def.machine.spls.maxiter = 100;
else
% Tolerance
def.machine.svd.tol = 1e-10;
% Explained variance
if strcmp(cfg.machine.name, 'cca')
def.machine.svd.varx = 1; % Note that it can be effectively overwritten by hyperparameter
def.machine.svd.vary = 1;
elseif strcmp(cfg.machine.name, 'rcca')
def.machine.svd.varx = 0.99;
def.machine.svd.vary = 0.99; % No hyperparameter here, so it always has an effect
end
end
% Update fields
cfg = assign_defaults(cfg, def);
% Sanity checks for hyperparameter settings
for i=1:numel(cfg.data.mod)
m = lower(cfg.data.mod{i}); % shorthand for modality in lowercase
% Check if L1 regularization obeys limits
if strcmp(cfg.machine.name, 'spls') && (cfg.machine.param.(['rangeL1' m])(1) < 1 || ...
cfg.machine.param.(['rangeL1' m])(2) > sqrt(cfg.data.(upper(m)).nfeat))
error('L1 regularization is out of interval: [1, sqrt(size(%s, 2)].', upper(m));
end
end
% Check and update machine parameter fields
fields = fieldnames(cfg.machine.param);
for i=1:numel(fields)
if ~isempty(regexp(fields{i}, 'range')) || ~isempty(regexp(fields{i}, 'n.*(x|y)'))
try
def.machine.param = rmfield(def.machine.param, fields{i});
catch; end
cfg.machine.param = rmfield(cfg.machine.param, fields{i});
end
end
% Check if grid search needed
param = get_hyperparam(cfg, 'default');
% Check and update framework settings
if numel(param) == 1
if isfield(def.frwork.split, 'nin')
def.frwork.split = rmfield(def.frwork.split, 'nin');
end
if isfield(cfg.frwork.split, 'nin')
cfg.frwork.split = rmfield(cfg.frwork.split, 'nin');
end
if isfield(def.frwork.split, 'propin')
def.frwork.split = rmfield(def.frwork.split, 'propin');
end
if isfield(cfg.frwork.split, 'propin')
cfg.frwork.split = rmfield(cfg.frwork.split, 'propin');
end
end
% Project subdirectories
switch cfg.frwork.name
case 'holdout'
if numel(param) > 1
def.dir.frwork = sprintf('%s_holdout%d-%.2f_subsamp%d-%.2f', cfg.machine.name, ...
cfg.frwork.split.nout, cfg.frwork.split.propout, cfg.frwork.split.nin, cfg.frwork.split.propin);
else
def.dir.frwork = sprintf('%s_holdout%d-%.2f', cfg.machine.name, cfg.frwork.split.nout, ...
cfg.frwork.split.propout);
end
case 'permutation'
def.dir.frwork = sprintf('%s_permutation', cfg.machine.name);
end
if all(isfield(cfg.machine.param, {'PCAx' 'PCAy'}))
if numel(cfg.machine.param.PCAx) == 1 && numel(cfg.machine.param.PCAy) == 1
if cfg.machine.param.PCAx == cfg.machine.param.PCAy
def.dir.frwork = strrep(def.dir.frwork, 'cca', sprintf('cca_pca%d', cfg.machine.param.PCAx));
else
def.dir.frwork = strrep(def.dir.frwork, 'cca', sprintf('cca_pca%d-%d', ...
cfg.machine.param.PCAx, cfg.machine.param.PCAy));
end
else
def.dir.frwork = strrep(def.dir.frwork, 'cca', 'cca_pca');
end
end
def.dir.frwork = [fullfile(cfg.dir.project, 'framework', def.dir.frwork) cfg.frwork.flag];
def.dir.load = fullfile(def.dir.frwork, 'load');
% Update fields
cfg = assign_defaults(cfg, def);
% Check if statistical inference is supported
if cfg.stat.nperm > 0
if strcmp(cfg.machine.name, 'pls') && ~ismember(cfg.stat.crit, {'correl' 'covar'})
error('Statistical inference should be based on correlation or covariance.')
elseif ~strcmp(cfg.stat.crit, 'correl')
error('Statistical inference should be based on correlation.')
end
end
%----- Save cfg
if ~isdir(cfg.dir.frwork)
mkdir(cfg.dir.frwork)
end
savemat(cfg, fullfile(cfg.dir.frwork, 'cfg.mat'), 'cfg', cfg);

cfg_defaults.m at commit 6e1ef1e, under GPL-3.0 · at the source

Overview

Authors: Wenrui Bao^1,2, Gengchen Ye¹, Tao Wang^3,4,5, Xingpu Quan¹, Qiange Zhu¹, Liming Fan⁶, Haining Li⁷, Wentao Zeng¹, Junya Mu¹, Ruiting Zhu^1,2, Jixin Liu⁸, Yuchen Zhang⁹, Xuan Niu¹

ORCID iDs: Liming Fan, Xuan Niu

Department of Medical Imaging, the First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, Shaanxi Province 710061, China
School of Future Technology, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710049, China
Key Laboratory of Shaanxi Province for Craniofacial Precision Medicine Research, College of Stomatology, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710004, China
Laboratory Center of Stomatology, College of Stomatology, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710004, China
Department of Medical Imaging, College of Stomatology, Xi’an Jiaotong University, Xi'an, Shaanxi Province 710004, China
The Key Laboratory of Biomedical Information Engineering of Ministry of Education, Institute of Health and Rehabilitation Science, Xi’an Jiaotong University School of Life Science and Technology, Xi’an, Shaanxi Province 710049, China
PET/CT Center, the First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, Shaanxi Province 710061, China
School of Life Science and Technology, Xidian University, Xi'an Key Laboratory of Intelligent Sensing and Regulation of Trans-Scale Life Information, Xi’an, Shaanxi 710126, China
Department of Nuclear Medicine, the First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, Shaanxi Province 710061, China

Institutions: First Affiliated Hospital of Xi'an Jiaotong University (China); Xi'an Jiaotong University (China); Xidian University (China)

Journal: Brain communications, volume 8, issue 4, article fcag232

Dates: received 16 December 2025; accepted 27 May 2026; published online 18 June 2026

Type: Research article · Language: English

License: CC BY

Identifiers: DOI 10.1093/braincomms/fcag232 · PMID 42422532 · PMCID PMC13343389 · OpenAlex W7165172786

Open access: gold, a free copy (OpenAlex)

Status: code verified

Categories: structural MRI / diffusion (modality), human (organism), other condition (population), clinical / translational (subfield)

Methods: Statistics, Machine learning, fMRI & imaging, Smoothing, state filtering, decompositions, Preprocessing

Keywords: long COVID, neuropsychiatric symptoms, cortical thickness, grey matter volume, regularized canonical correlation analysis

Topic: Long-Term Effects of COVID-19 (Neurology, Medicine), according to OpenAlex

Funding: National Natural Science Foundation of China (National Science Foundation of China) (82202121); Health Research and Innovation Capacity Strengthening Platform Program of Shaanxi Province (2023PT-09); Natural Science Foundation of Shaanxi Province (2024JC-YBQN-0830); Chongqing Science and Technology Innovation Leading Talent Support Program (0959202513036); Shaanxi Provincial Young Star of Science and Technology Program (2025ZC-KJXX-125)

Citations: not cited yet (Europe PMC); 65 references in the paper

Abstract

Long COVID presents with heterogeneous and persistent neuropsychiatric symptoms, suggesting possible shared neural underpinnings. However, the dynamic brain structure–symptom relationships and their potential for predicting long-term outcomes remain unclear. We conducted a longitudinal study of 144 individuals with long COVID (mean age: 37.8 ± 10.1 years, 48.6% male). All participants experienced mild COVID-19 and were not hospitalized. Structural MRI and comprehensive psychiatric and cognitive assessments were performed at 1, 2 and 12 months post-infection. A cohort of 68 healthy controls (mean age: 36.0 ± 10.3 years; 41.2% male) completed the same assessment protocol at baseline. Regularized canonical correlation analysis was employed to identify multivariate associations between 13 psychiatric and cognitive measures and both grey matter volume and cortical thickness, and to assess whether early structural features predicted symptom outcomes at 1 year. Neuropsychiatric symptoms were linked to coordinated structural covariance patterns across distributed brain regions in long COVID; these associations strengthened from 1 to 3 months post-infection and were absent in controls. A stable cognitive-affective symptom dimension showed the strongest brain–behaviour coupling. Notably, the neural substrates of this coupling diverged over time: grey matter volume associations remained localized to prefrontal-limbic circuits, whereas cortical thickness associations expanded to frontoparietal regions. Critically, reduced grey matter volume in the right cuneus and superior frontal gyri, along with decreased cortical thickness in the left supramarginal gyrus at 3 months post-infection, were significantly linked to poorer executive function and greater fatigue 1 year later. Our findings delineate evolving neuroanatomical signature of long COVID, where distinct patterns of grey matter volume and cortical thickness underpin a core symptom profile and predict long-term neuropsychiatric symptoms. These results provide insight into the neural mechanisms of long COVID and identify specific targets for monitoring and early intervention.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above.

anaston/cca_pls_toolkit

License: GPL-3.0

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Commit: 6e1ef1e04c58984b1e9e8aa22dd5aabe490fb4de, 21 November 2022

Languages: MATLAB (72), JavaScript (28), Python (2)

Size: 275 files, 102 scripts

Software Heritage: archived

Found in: the text, “Regularized canonical correlation analysis”

Holds: README, license file, tests, continuous integration, documentation

Not found: CITATION.cff, environment file

Tools: Statistics and Machine Learning Toolbox (13 files), SPM (4 files)

Availability: 1 check, the latest on 27 September 2026: the link answers

27 September 2026: the link answers

104 files

cfg_defaults.m, MATLAB, 520 lines
demo/demo_simul_paper.m, MATLAB, 115 lines
documentation/matdoc.py, Python, 238 lines
documentation/matdocparser.py, Python, 383 lines
documentation/site/assets/javascripts/bundle.17f42bbf.min.js, JavaScript, 29 lines
documentation/site/assets/javascripts/lunr/min/lunr.ar.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.da.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.de.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.du.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.es.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.fi.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.fr.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.hi.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.hu.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.it.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.ja.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.jp.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.multi.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.nl.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.no.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.pt.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.ro.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.ru.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.stemmer.support.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.sv.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.th.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.tr.min.js, JavaScript, 18 lines
documentation/site/assets/javascripts/lunr/min/lunr.vi.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/min/lunr.zh.min.js, JavaScript, 1 line
documentation/site/assets/javascripts/lunr/tinyseg.js, JavaScript, 206 lines
documentation/site/assets/javascripts/lunr/wordcut.js, JavaScript, 186 lines
documentation/site/assets/javascripts/workers/search.cefbb252.min.js, JavaScript, 48 lines
examples/example_fmri.m, MATLAB, 419 lines
examples/example_simulation.m, MATLAB, 375 lines
examples/example_smri.m, MATLAB, 406 lines
fileio/cleanup_files.m, MATLAB, 82 lines
fileio/compile_files.m, MATLAB, 74 lines
fileio/exist_file.m, MATLAB, 52 lines
fileio/getfname.m, MATLAB, 51 lines
fileio/loadmat.m, MATLAB, 63 lines
fileio/loadmat_struct.m, MATLAB, 56 lines
fileio/parse_input.m, MATLAB, 54 lines
fileio/parse_struct.m, MATLAB, 54 lines
fileio/renamemat.m, MATLAB, 63 lines
fileio/savemat.m, MATLAB, 76 lines
fileio/select_file.m, MATLAB, 37 lines
machines/rcca.m, MATLAB, 44 lines
machines/spls.m, MATLAB, 187 lines
misc/calc_permid.m, MATLAB, 107 lines
misc/calc_proj.m, MATLAB, 41 lines
misc/calc_splits.m, MATLAB, 124 lines
misc/concat_data.m, MATLAB, 48 lines
misc/deflate_data.m, MATLAB, 113 lines
misc/generate_data.m, MATLAB, 86 lines
misc/get_featid.m, MATLAB, 44 lines
misc/get_hyperparam.m, MATLAB, 97 lines
misc/impute_mat.m, MATLAB, 53 lines
misc/load_data.m, MATLAB, 359 lines
misc/main.m, MATLAB, 83 lines
misc/permute_data.m, MATLAB, 73 lines
misc/preproc_data.m, MATLAB, 123 lines
misc/process_metric.m, MATLAB, 68 lines
misc/qc_data.m, MATLAB, 85 lines
misc/run_machine.m, MATLAB, 112 lines
misc/run_model.m, MATLAB, 299 lines
misc/save_results.m, MATLAB, 149 lines
misc/stat_inference.m, MATLAB, 74 lines
plot/plot_paropt.m, MATLAB, 207 lines
plot/plot_proj.m, MATLAB, 309 lines
plot/plot_proj_2d.m, MATLAB, 62 lines
plot/plot_proj_2d_cmap.m, MATLAB, 64 lines
plot/plot_proj_2d_group.m, MATLAB, 95 lines
plot/plot_weight.m, MATLAB, 219 lines
plot/plot_weight_behav_horz.m, MATLAB, 116 lines
plot/plot_weight_behav_text.m, MATLAB, 146 lines
plot/plot_weight_behav_vert.m, MATLAB, 107 lines
plot/plot_weight_brain_conn_node.m, MATLAB, 84 lines
plot/plot_weight_brain_cortex.m, MATLAB, 65 lines
plot/plot_weight_brain_edge.m, MATLAB, 98 lines
plot/plot_weight_brain_module.m, MATLAB, 130 lines
plot/plot_weight_brain_node.m, MATLAB, 65 lines
plot/plot_weight_stem.m, MATLAB, 99 lines
res_defaults.m, MATLAB, 314 lines
set_path.m, MATLAB, 100 lines
test/get_data.m, MATLAB, 56 lines
test/test_algorithms.m, MATLAB, 165 lines
test/test_pipeline.m, MATLAB, 96 lines
util/assign_defaults.m, MATLAB, 42 lines
util/calc_distance.m, MATLAB, 32 lines
util/calc_exvar.m, MATLAB, 113 lines
util/calc_stability.m, MATLAB, 130 lines
util/cov2.m, MATLAB, 37 lines
util/deflation.m, MATLAB, 108 lines
util/fastsvd.m, MATLAB, 139 lines
util/ffd_val_str.m, MATLAB, 50 lines
util/fix_fileend.m, MATLAB, 58 lines
util/init_brainnet.m, MATLAB, 63 lines
util/mean_center_features.m, MATLAB, 36 lines
util/norm_features.m, MATLAB, 62 lines
util/normalize2MNI.m, MATLAB, 77 lines
util/postproc_weight.m, MATLAB, 57 lines
util/update_dir.m, MATLAB, 135 lines
LICENSE, License, 692 lines
README.md, Text, 49 lines

At the source: github.com/anaston/cca_pls_toolkit

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
102 scripts, each with its path and the digest of its content;
no match between paragraphs and code yet;
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data availability

To get access to the data and comply with the terms of our research ethics committee approval, an application to the corresponding author will be required, specifying the geographical extent of sharing.

The regularized canonical correlation analysis (RCCA) was performed using the publicly available CCA/PLS toolkit developed by the Machine Learning & Neuroimaging Laboratory, University College London (https://github.com/anaston/cca_pls_toolkit), which is openly accessible.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 13 authors, 5 keywords, 5 funders, 65 references.

Cite

This paper

Bao, W., Ye, G., Wang, T., Quan, X., Zhu, Q., Fan, L., Li, H., Zeng, W., Mu, J., Zhu, R., Liu, J., Zhang, Y., & Niu, X. (2026). Dynamic multivariate patterns of brain structure-neuropsychiatric symptom associations in long COVID. Brain communications, 8(4), fcag232. https://doi.org/10.1093/braincomms/fcag232

BibTeX

@article{bao2026dynamic, author = {Bao, Wenrui and Ye, Gengchen and Wang, Tao and Quan, Xingpu and Zhu, Qiange and Fan, Liming and Li, Haining and Zeng, Wentao and Mu, Junya and Zhu, Ruiting and Liu, Jixin and Zhang, Yuchen and Niu, Xuan}, title = {{Dynamic multivariate patterns of brain structure-neuropsychiatric symptom associations in long COVID}}, journal = {Brain communications}, year = {2026}, month = jun, volume = {8}, number = {4}, pages = {fcag232}, publisher = {Oxford University Press}, issn = {2632-1297}, doi = {10.1093/braincomms/fcag232}, url = {https://doi.org/10.1093/braincomms/fcag232}, pmid = {42422532}, pmcid = {PMC13343389} }

RIS

TY - JOUR AU - Bao, Wenrui AU - Ye, Gengchen AU - Wang, Tao AU - Quan, Xingpu AU - Zhu, Qiange AU - Fan, Liming AU - Li, Haining AU - Zeng, Wentao AU - Mu, Junya AU - Zhu, Ruiting AU - Liu, Jixin AU - Zhang, Yuchen AU - Niu, Xuan TI - Dynamic multivariate patterns of brain structure-neuropsychiatric symptom associations in long COVID T2 - Brain communications J2 - Brain Commun PY - 2026 DA - 2026/06/18 VL - 8 IS - 4 SP - fcag232 SN - 2632-1297 PB - Oxford University Press DO - 10.1093/braincomms/fcag232 UR - https://doi.org/10.1093/braincomms/fcag232 LA - en ER -

CSL-JSON

{ "id": "10.1093/braincomms/fcag232", "type": "article-journal", "title": "Dynamic multivariate patterns of brain structure-neuropsychiatric symptom associations in long COVID", "container-title": "Brain communications", "author": [ { "family": "Bao", "given": "Wenrui" }, { "family": "Ye", "given": "Gengchen" }, { "family": "Wang", "given": "Tao" }, { "family": "Quan", "given": "Xingpu" }, { "family": "Zhu", "given": "Qiange" }, { "family": "Fan", "given": "Liming" }, { "family": "Li", "given": "Haining" }, { "family": "Zeng", "given": "Wentao" }, { "family": "Mu", "given": "Junya" }, { "family": "Zhu", "given": "Ruiting" }, { "family": "Liu", "given": "Jixin" }, { "family": "Zhang", "given": "Yuchen" }, { "family": "Niu", "given": "Xuan" } ], "container-title-short": "Brain Commun", "volume": "8", "issue": "4", "page": "fcag232", "DOI": "10.1093/braincomms/fcag232", "PMID": "42422532", "PMCID": "PMC13343389", "ISSN": "2632-1297", "publisher": "Oxford University Press", "URL": "https://doi.org/10.1093/braincomms/fcag232", "language": "en", "issued": { "date-parts": [ [ 2026, 6, 18 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1016/j.bbih.2026.101274 [code]: Brain dynamics of attentional, default-mode and limbic networks are disrupted at rest in post-COVID-19 syndrome.
Journal: Brain, behavior, & immunity - health
In common: SPM, Statistics and Machine Learning Toolbox, other condition, 3 references
[2] doi:10.1007/s11682-026-01175-1: Longitudinal surface-based morphometry reveals potential MRI biomarkers correlated with multidimensional brain health measures.
Journal: Brain imaging and behavior
In common: structural MRI / diffusion, clinical / translational, 3 references
[3] doi:10.3389/fimmu.2026.1776555 [code]: Multi-omics analysis of long COVID (post-COVID-19 condition) reveals persistent mitochondrial dysfunction, suppressed oxidative phosphorylation, and immune dysregulation.
Journal: Frontiers in immunology
In common: other condition, 3 references
[4] doi:10.3390/brainsci16040410: LearningRx Cognitive Training for Workplace Self-Efficacy in Adults with Post-COVID-19 Brain Fog: A Mixed-Methods Pilot Study.
Journal: Brain sciences
In common: other condition, 3 references
[5] doi:10.1038/s41467-026-74099-5 [code]: Right amygdala ablation reduces maladaptive negative interpretation bias and symptoms in a patient with post-traumatic stress disorder.
Journal: Nature communications
In common: SPM, Statistics and Machine Learning Toolbox, clinical / translational, other condition, 1 reference
[6] doi:10.1038/s41467-026-76156-5 [code]: Longitudinal analysis reveals myeloid cell contributions to murine neuroPASC pathogenesis.
Journal: Nature communications
In common: other condition, 3 references
[7] doi:10.1093/braincomms/fcag045 [code]: Examining neuroimaging biomarkers, plasma biomarkers and cognitive functions in patients with recovered COVID-19 infection: a multicentre study using 7T MRI.
Journal: Brain communications
In common: structural MRI / diffusion, clinical / translational, other condition, 2 references
[8] doi:10.1016/j.bbih.2026.101299 [code]: Multimodal approach to identify neuropsychophysiological subgroups in myalgic encephalomyelitis/chronic fatigue syndrome and their relevance for rehabilitation: protocol for a mechanistic cross-sectional and longitudinal study.
Journal: Brain, behavior, & immunity - health
In common: SPM, Statistics and Machine Learning Toolbox, clinical / translational, 1 reference
[9] doi:10.1038/s41597-026-07676-4 [code]: A neuroimaging dataset combining movie-watching, eye-tracking, sensorimotor mapping, and cognitive tasks.
Journal: Scientific data
In common: SPM, Statistics and Machine Learning Toolbox, structural MRI / diffusion, 1 reference
[10] doi:10.1038/s41593-026-02359-0 [code]: The cross-site reproducibility of MRI morphometric phenotypes in psychiatric disorders.
Journal: Nature neuroscience
In common: SPM, Statistics and Machine Learning Toolbox, structural MRI / diffusion, 1 reference

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.