OSCR

Resting-state brain age is associated with cognitive and sensorimotor abnormalities in schizophrenia spectrum disorders: a validation & longitudinal study.

Code ↔ Paper

1 match between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

One block of code here has a permanent identifier (oscr:b:…), listed with the matches below: citable for good, with a page of its own. Select a match and press Cite to copy it. What a block identifier is.

The 1 match
  1. [1] § Methods › Site harmonization & confounding ↔ neuroHarmonize/harmonizationLearn.py, lines 10 to 78 · score 0.61 · linear model, ComBat, harmonization model, residuals, covariates · block oscr:b:K142D45QRCHDH-7 · Report a mismatch

Paper

Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 407 lines · 19 KB · MIT · 1 match

  1. import inspect
  2. import os
  3. import pickle
  4. import numpy as np
  5. import pandas as pd
  6. from statsmodels.gam.api import GLMGam, BSplines
  7. from .harmonizationApply import applyStandardizationAcrossFeatures
  8. from neuroCombat.neuroCombat import make_design_matrix, find_parametric_adjustments, adjust_data_final, aprior, bprior
  9. import copy
  10. def harmonizationLearn(data, covars, eb=True, smooth_terms=[], smooth_term_bounds=(None, None),
  11. ref_batch=None, return_s_data=False,
  12. orig_model=None, seed=None):
  13. """
  14. Wrapper for neuroCombat function that returns the harmonization model.
  15. Arguments
  16. ---------
  17. data : a numpy array
  18. data to harmonize with ComBat, dimensions are N_samples x N_features
  19. covars : a pandas DataFrame
  20. contains covariates to control for during harmonization
  21. all covariates must be encoded numerically (no categorical variables)
  22. must contain a single column "SITE" with site labels for ComBat
  23. dimensions are N_samples x (N_covariates + 1)
  24. eb : bool, default True
  25. whether to use empirical Bayes estimates of site effects
  26. smooth_terms (Optional) : a list, default []
  27. names of columns in covars to include as smooth, nonlinear terms
  28. can be any or all columns in covars, except "SITE"
  29. if empty, ComBat is applied with a linear model of covariates
  30. if not empty, Generalized Additive Models (GAMs) are used
  31. will increase computation time due to search for optimal smoothing
  32. smooth_term_bounds (Optional) : tuple of float, default (None, None)
  33. feature to support custom boundaries of the smoothing terms
  34. useful when holdout data covers different range than
  35. specify the bounds as (minimum, maximum)
  36. currently not supported for models with mutliple smooth terms
  37. ref_batch (Optional) : str or int, default None
  38. batch (site or scanner) to be used as reference for batch adjustment
  39. return_s_data (Optional) : bool, default False
  40. whether to return s_data, the standardized data array
  41. can be useful for diagnostics but will be costly to save/load if large
  42. seed (Optional) : int, default None
  43. By default, this function is non-deterministic. Setting the optional
  44. argument `seed` will make the function deterministic.
  45. Returns
  46. -------
  47. model : a dictionary of estimated model parameters
  48. design, var_pooled, B_hat, stand_mean, mod_mean,
  49. gamma_star, delta_star, info_dict (a neuroCombat invention),
  50. gamma_hat, delta_hat, gamma_bar, t2, a_prior, b_prior, smooth_model
  51. bayes_data : a numpy array
  52. harmonized data, corrected for effects of SITE
  53. dimensions are N_samples x N_features
  54. s_data (Optional) : a numpy array
  55. standardized residuals after accounting for `covars` other than `SITE`
  56. set return_s_data=True to output the variable
  57. """
  58. # set optional random seed
  59. if seed is not None:
  60. np.random.seed(seed)
  61. rng = np.random.default_rng(seed) if seed is not None else None
  62. if orig_model is None:
  63. pass
  64. else:
  65. model = copy.deepcopy(orig_model)
  66. # bypass EB step if only one variable provided in dataset
  67. if data.shape[1]==1:
  68. if eb:
  69. print('\n[neuroHarmonize]: Bypassing empirical Bayes step because only one variable to harmonize.')
  70. eb = False
  71. # transpose data as per ComBat convention
  72. data = data.T
  73. # prep covariate data
  74. covar_levels = list(covars.columns)
  75. batch_labels = np.unique(covars.SITE)
  76. batch_col = covars.columns.get_loc('SITE')
  77. if orig_model is None:
  78. pass
  79. else:
  80. isTrainSite = covars['SITE'].isin(model['SITE_labels'])
  81. isTrainSiteLabel = set(model['SITE_labels'])
  82. isTrainSiteColumns = np.where((pd.DataFrame(np.unique(covars['SITE'])).isin(model['SITE_labels']).values).flat)
  83. isTrainSiteColumnsOrig = np.where((pd.DataFrame(model['SITE_labels']).isin(np.unique(covars['SITE'])).values).flat)
  84. isTestSiteColumns = np.where((~pd.DataFrame(np.unique(covars['SITE'])).isin(model['SITE_labels']).values).flat)
  85. cat_cols = []
  86. num_cols = [covars.columns.get_loc(c) for c in covars.columns if c!='SITE']
  87. smooth_cols = [covars.columns.get_loc(c) for c in covars.columns if c in smooth_terms]
  88. # maintain a dictionary of smoothing information
  89. smooth_model = {
  90. 'perform_smoothing': len(smooth_terms)>0,
  91. 'smooth_terms': smooth_terms,
  92. 'smooth_cols': smooth_cols,
  93. 'bsplines_constructor': None,
  94. 'formula': None,
  95. 'df_gam': None
  96. }
  97. covars = np.array(covars, dtype='object')
  98. if ref_batch is None:
  99. covars[:,batch_col] = np.unique(covars[:,batch_col],return_inverse=True)[-1]
  100. ref_level=None
  101. else:
  102. ref_indices = np.argwhere((covars[:,batch_col]==ref_batch).squeeze())
  103. covars[:,batch_col] = np.unique(covars[:,batch_col],return_inverse=True)[-1]
  104. if ref_indices.shape[0]==0:
  105. ref_level=None
  106. ref_batch=None
  107. print('[neuroHarmonize] batch.ref not found. Setting to None.')
  108. else:
  109. ref_level = int(covars[ref_indices[0],batch_col])
  110. # create dictionary that stores batch info
  111. (batch_levels, sample_per_batch) = np.unique(covars[:,batch_col],return_counts=True)
  112. info_dict = {
  113. 'batch_levels': batch_levels.astype('int'),
  114. 'n_batch': len(batch_levels),
  115. 'n_sample': int(covars.shape[0]),
  116. 'sample_per_batch': sample_per_batch.astype('int'),
  117. 'batch_info': [list(np.where(covars[:,batch_col]==idx)[0]) for idx in batch_levels],
  118. 'ref_level': ref_level
  119. }
  120. ###
  121. design = make_design_matrix(covars, batch_col, cat_cols, num_cols, ref_level)
  122. ### additional setup if smoothing is performed
  123. if smooth_model['perform_smoothing']:
  124. # create cubic spline basis for smooth terms
  125. X_spline = covars[:, smooth_cols].astype(float)
  126. if orig_model is None:
  127. if len(smooth_cols)==1:
  128. bs = BSplines(X_spline, df=10, degree=3,
  129. knot_kwds=[{'lower_bound':smooth_term_bounds[0], 'upper_bound':smooth_term_bounds[1]}])
  130. else:
  131. bs = BSplines(X_spline, df=[10] * len(smooth_cols), degree=[3] * len(smooth_cols))
  132. # construct formula and dataframe required for gam
  133. formula = 'y ~ '
  134. df_gam = {}
  135. for b in batch_levels:
  136. formula = formula + 'x' + str(b) + ' + '
  137. df_gam['x' + str(b)] = design[:, b]
  138. for c in num_cols:
  139. if c not in smooth_cols:
  140. formula = formula + 'c' + str(c) + ' + '
  141. df_gam['c' + str(c)] = covars[:, c].astype(float)
  142. formula = formula[:-2] + '- 1'
  143. df_gam = pd.DataFrame(df_gam)
  144. # for matrix operations, a modified design matrix is required
  145. design = np.concatenate((df_gam, bs.basis), axis=1)
  146. # store objects in dictionary
  147. smooth_model['bsplines_constructor'] = bs
  148. smooth_model['formula'] = formula
  149. smooth_model['df_gam'] = df_gam
  150. else:
  151. bs_basis = model['smooth_model']['bsplines_constructor'].transform(X_spline)
  152. # construct formula and dataframe required for gam
  153. formula = 'y ~ '
  154. df_gam = {}
  155. for b in batch_levels:
  156. formula = formula + 'x' + str(b) + ' + '
  157. df_gam['x' + str(b)] = design[:, b]
  158. for c in num_cols:
  159. if c not in smooth_cols:
  160. formula = formula + 'c' + str(c) + ' + '
  161. df_gam['c' + str(c)] = covars[:, c].astype(float)
  162. formula = formula[:-2] + '- 1'
  163. df_gam = pd.DataFrame(df_gam)
  164. # for matrix operations, a modified design matrix is required
  165. design = np.concatenate((df_gam, bs_basis), axis=1)
  166. ###
  167. # run steps to perform ComBat
  168. if orig_model is None:
  169. s_data, stand_mean, var_pooled, mod_mean, B_hat = standardizeAcrossFeatures(
  170. data, design, info_dict, smooth_model, rng=rng)
  171. LS_dict = fitLSModelAndFindPriors(s_data, design, info_dict, eb=eb)
  172. # optional: avoid EB estimates
  173. if eb:
  174. gamma_star, delta_star = find_parametric_adjustments(s_data, LS_dict, info_dict, mean_only=False)
  175. else:
  176. gamma_star = LS_dict['gamma_hat']
  177. delta_star = np.array(LS_dict['delta_hat'])
  178. bayes_data = adjust_data_final(s_data, design, gamma_star, delta_star, stand_mean, mod_mean, var_pooled, info_dict, data)
  179. # save model parameters in single object
  180. model = {'design': design, 'SITE_labels': batch_labels,
  181. 'var_pooled':var_pooled, 'B_hat':B_hat, 'stand_mean': stand_mean, 'mod_mean': mod_mean,
  182. 'gamma_star': gamma_star, 'delta_star': delta_star, 'info_dict': info_dict,
  183. 'gamma_hat': LS_dict['gamma_hat'], 'delta_hat': np.array(LS_dict['delta_hat']),
  184. 'gamma_bar': LS_dict['gamma_bar'], 't2': LS_dict['t2'],
  185. 'a_prior': LS_dict['a_prior'], 'b_prior': LS_dict['b_prior'],
  186. 'smooth_model': smooth_model, 'eb': eb,'SITE_labels_train':batch_labels,'Covariates':covar_levels,
  187. 'ref_batch': ref_batch}
  188. # transpose data to return to original shape
  189. bayes_data = bayes_data.T
  190. else:
  191. # Create train data
  192. (batch_levels, sample_per_batch) = np.unique(covars[isTrainSite,batch_col],return_counts=True)
  193. if batch_levels.size == 0:
  194. bayes_data_train = np.zeros(shape=(0,data.shape[0]))
  195. s_data_train = np.zeros(shape=(0,data.shape[0])).T
  196. else:
  197. info_dict_train = model['info_dict'].copy()
  198. info_dict_train['sample_per_batch'] = sample_per_batch.astype('int')
  199. info_dict_train['batch_info'] = [list(np.where(covars[isTrainSite,batch_col]==idx)[0]) for idx in batch_levels]
  200. tmp = np.concatenate((np.zeros(shape=(info_dict['n_sample'],len(model['SITE_labels']))), design[:,len(batch_labels):]),axis=1)
  201. s_data_train, stand_mean_train, var_pooled_train, mod_mean_train = applyStandardizationAcrossFeatures(data[:,isTrainSite], tmp[isTrainSite,:], info_dict_train, model)
  202. design2=tmp.copy()
  203. design2[:,isTrainSiteColumnsOrig[0]] = design[:,isTrainSiteColumns[0]]
  204. bayes_data_train = adjust_data_final(s_data_train, design2[isTrainSite,:], model['gamma_star'], model['delta_star'], stand_mean_train, mod_mean_train, model['var_pooled'], info_dict_train, data)
  205. # transpose data to return to original shape
  206. bayes_data_train = bayes_data_train.T
  207. # Create test data (new SITE)
  208. (batch_levels, sample_per_batch) = np.unique(covars[~isTrainSite,batch_col],return_counts=True)
  209. if batch_levels.size == 0:
  210. bayes_data_test = np.zeros(shape=(0,data.shape[0]))
  211. s_data_test = np.zeros(shape=(0,data.shape[0])).T
  212. else:
  213. info_dict_test = {
  214. 'batch_levels': batch_levels.astype('int'),
  215. 'n_batch': len(batch_levels),
  216. 'n_sample': int(covars[~isTrainSite,:].shape[0]),
  217. 'sample_per_batch': sample_per_batch.astype('int'),
  218. 'batch_info': [list(np.where(covars[~isTrainSite,batch_col]==idx)[0]) for idx in batch_levels]
  219. }
  220. design_tmp = np.concatenate((design[:,isTestSiteColumns[0]], design[:,len(batch_labels):]),axis=1)
  221. s_data_test, stand_mean_test, var_pooled_test, mod_mean_test = applyStandardizationAcrossFeatures(data[:,~isTrainSite], design_tmp[~isTrainSite,:], info_dict_test, model)
  222. LS_dict = fitLSModelAndFindPriors(s_data_test, design_tmp[~isTrainSite,:], info_dict_test, eb=eb)
  223. if eb:
  224. gamma_star, delta_star = find_parametric_adjustments(s_data_test, LS_dict, info_dict_test, mean_only=False)
  225. else:
  226. gamma_star = LS_dict['gamma_hat']
  227. delta_star = np.array(LS_dict['delta_hat'])
  228. betas = []
  229. for i in range(info_dict_test['n_batch']):
  230. diff_mean = np.mean(data[:,info_dict_test['batch_info'][i]]-np.dot(design[info_dict_test['batch_info'][i],info_dict['n_batch']:],model['B_hat'][len(model['SITE_labels']):,:]).T,axis=1)
  231. betas.append(diff_mean)
  232. new_betas = np.array(betas)
  233. model['B_hat'] = np.concatenate((model['B_hat'][:len(model['SITE_labels']),:],new_betas,model['B_hat'][len(model['SITE_labels']):,:]))
  234. model['SITE_labels'] = np.append(model['SITE_labels'],list(set(batch_labels)-isTrainSiteLabel))
  235. model['gamma_star'] = np.append(model['gamma_star'],gamma_star,axis=0)
  236. model['delta_star'] = np.append(model['delta_star'],delta_star,axis=0)
  237. model['info_dict']['n_batch'] = len(model['SITE_labels'])
  238. bayes_data_test = adjust_data_final(s_data_test, design_tmp[~isTrainSite,:], gamma_star, delta_star, stand_mean_test, mod_mean_test, model['var_pooled'], info_dict_test, data)
  239. # transpose data to return to original shape
  240. bayes_data_test = bayes_data_test.T
  241. bayes_data = np.zeros(shape=data.T.shape)
  242. bayes_data[isTrainSite,:] = bayes_data_train
  243. bayes_data[~isTrainSite,:] = bayes_data_test
  244. s_data = np.zeros(shape=data.T.shape)
  245. s_data[isTrainSite,:] = s_data_train.T
  246. s_data[~isTrainSite,:] = s_data_test.T
  247. if return_s_data:
  248. return model, bayes_data, s_data.T
  249. else:
  250. return model, bayes_data
  251. def standardizeAcrossFeatures(X, design, info_dict, smooth_model, rng=None):
  252. """
  253. The original neuroCombat function standardize_across_features plus
  254. necessary modifications.
  255. This function will return all estimated parameters in addition to the
  256. standardized data.
  257. """
  258. n_batch = info_dict['n_batch']
  259. n_sample = info_dict['n_sample']
  260. sample_per_batch = info_dict['sample_per_batch']
  261. ### perform smoothing with GAMs if specified
  262. if smooth_model['perform_smoothing']:
  263. smooth_cols = smooth_model['smooth_cols']
  264. bs = smooth_model['bsplines_constructor']
  265. formula = smooth_model['formula']
  266. df_gam = smooth_model['df_gam']
  267. if X.shape[0] > 10:
  268. print('\n[neuroHarmonize]: smoothing more than 10 variables may take several minutes of computation.')
  269. # initialize penalization weight (not the final weight)
  270. alpha = np.array([1.0] * len(smooth_cols))
  271. # initialize an empty matrix for beta
  272. B_hat = np.zeros((design.shape[1], X.shape[0]))
  273. # estimate beta for each variable to be harmonized
  274. for i in range(0, X.shape[0]):
  275. df_gam.loc[:, 'y'] = X[i, :]
  276. gam_bs = GLMGam.from_formula(formula, data=df_gam, smoother=bs, alpha=alpha)
  277. res_bs = gam_bs.fit()
  278. # Optimal penalization weights alpha can be obtained through gcv/kfold
  279. # Note: kfold is faster, gcv is more robust
  280. select_penweight = gam_bs.select_penweight_kfold
  281. if rng is not None and 'rng' in inspect.signature(select_penweight).parameters:
  282. gam_bs.alpha = select_penweight(rng=rng)[0]
  283. else:
  284. gam_bs.alpha = select_penweight()[0]
  285. res_bs_optim = gam_bs.fit()
  286. B_hat[:, i] = res_bs_optim.params
  287. ###
  288. else:
  289. B_hat = np.dot(np.dot(np.linalg.inv(np.dot(design.T, design)), design.T), X.T)
  290. batch_info = info_dict['batch_info']
  291. ref_level = info_dict['ref_level']
  292. if ref_level is not None:
  293. grand_mean = np.transpose(B_hat[ref_level,:])
  294. X_ref = X[:,batch_info[ref_level]]
  295. design_ref = design[batch_info[ref_level],:]
  296. n_sample_ref = sample_per_batch[ref_level]
  297. var_pooled = np.dot(((X_ref - np.dot(design_ref, B_hat).T)**2), np.ones((n_sample_ref, 1)) / float(n_sample_ref))
  298. else:
  299. grand_mean = np.dot((sample_per_batch/ float(n_sample)).T, B_hat[:n_batch,:])
  300. var_pooled = np.dot(((X - np.dot(design, B_hat).T)**2), np.ones((n_sample, 1)) / float(n_sample))
  301. stand_mean = np.dot(grand_mean.T.reshape((len(grand_mean), 1)), np.ones((1, n_sample))) # nothing but grand mean
  302. # new code in neuroCombat to compute model mean
  303. if design is not None:
  304. tmp = copy.deepcopy(design)
  305. tmp[:,range(0,n_batch)] = 0
  306. mod_mean = np.transpose(np.dot(tmp, B_hat))
  307. s_data = ((X- stand_mean - mod_mean) / np.dot(np.sqrt(var_pooled), np.ones((1, n_sample))))
  308. return s_data, stand_mean, var_pooled, mod_mean, B_hat
  309. def fitLSModelAndFindPriors(s_data, design, info_dict, eb=True):
  310. """
  311. The original neuroCombat function fit_LS_model_and_find_priors plus
  312. necessary modifications.
  313. This function will return no EB information if eb=False
  314. """
  315. n_batch = info_dict['n_batch']
  316. batch_info = info_dict['batch_info']
  317. batch_design = design[:,:n_batch]
  318. gamma_hat = np.array(np.dot(np.dot(np.linalg.inv(np.matrix(np.dot(batch_design.T, batch_design))), batch_design.T), s_data.T))
  319. delta_hat = []
  320. for i, batch_idxs in enumerate(batch_info):
  321. delta_hat.append(np.var(s_data[:,batch_idxs],axis=1,ddof=1))
  322. if eb:
  323. gamma_bar = np.mean(gamma_hat, axis=1)
  324. t2 = np.var(gamma_hat,axis=1, ddof=1)
  325. a_prior = list(map(aprior, delta_hat))
  326. b_prior = list(map(bprior, delta_hat))
  327. LS_dict = {}
  328. LS_dict['gamma_hat'] = gamma_hat
  329. LS_dict['delta_hat'] = delta_hat
  330. LS_dict['gamma_bar'] = gamma_bar
  331. LS_dict['t2'] = t2
  332. LS_dict['a_prior'] = a_prior
  333. LS_dict['b_prior'] = b_prior
  334. return LS_dict
  335. else:
  336. LS_dict = {}
  337. LS_dict['gamma_hat'] = gamma_hat
  338. LS_dict['delta_hat'] = delta_hat
  339. LS_dict['gamma_bar'] = None
  340. LS_dict['t2'] = None
  341. LS_dict['a_prior'] = None
  342. LS_dict['b_prior'] = None
  343. return LS_dict
  344. def saveHarmonizationModel(model, file_name):
  345. """
  346. Save a harmonization model from harmonizationLearn().
  347. For saving model contents, this function will create a new file specified
  348. by file_name, and store the model using the pickle package.
  349. """
  350. if os.path.exists(file_name):
  351. raise ValueError('Model file already exists: %s. Change name or delete to save.' % file_name)
  352. # estimate size of out_file
  353. est_size = 0
  354. for key in ['design', 'B_hat', 'stand_mean', 'mod_mean', 'var_pooled',
  355. 'gamma_star', 'delta_star', 'gamma_hat', 'delta_hat']:
  356. est_size += model[key].nbytes / 1e6
  357. print('\n[neuroHarmonize]: Saving model object, estimated size in MB: %4.2f' % est_size)
  358. out_file = open(file_name, 'wb')
  359. pickle.dump(model, out_file)
  360. out_file.close()
  361. return None

harmonizationLearn.py at commit 22c550b, under MIT · at the source

Overview

Authors: Sebastian Volkmer1,2,3, Stefan Fritze1,3, Dilsa Cemre Akkoc Altinok1,3, Geva Brandt1, Ersoy Kocak1,2, Julius Wiegert1,2, Oksana Berhe1,3, Yuchen Lin1,3, Heike Tost1,3, Andreas Meyer-Lindenberg1,3, Dusan Hirjak1,3, Emanuel Schwarz1,2,3
  1. Department of Psychiatry and Psychotherapy, Central Institute of Mental Health, Medical Faculty Mannheim, University of Heidelberg,Mannheim, Germany
  2. Hector Institute for Artificial Intelligence in Psychiatry, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University,Mannheim, Germany
  3. German Centre for Mental Health (DZPG), Partner site Mannheim-Heidelberg-Ulm, Mannheim, Germany
Journal: Translational psychiatry, volume 16, issue 1, article 451
Dates: received 11 March 2026; accepted 25 August 2026; published online 4 September 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1038/s41398-026-04426-3 · PMID 42697895 · PMCID PMC13545253 · OpenAlex W7208788794
Open access: gold, a free copy (OpenAlex)
Status: code verified (how the code is found and verified)
Categories: structural MRI / diffusion (modality), human (organism), schizophrenia / psychosis (population)
Methods: Spectral & time-frequency, Statistics, Machine learning, Preprocessing, Connectivity, fMRI & imaging
Keywords: Diagnostic markers, Schizophrenia
MeSH: Aging*, Brain*, Schizophrenia*, Adolescent, Adult, Case-Control Studies, Cognition, Female, Humans, Longitudinal Studies, Magnetic Resonance Imaging, Male, Middle Aged, Young Adult (* major topic)
Topic: Functional Brain Connectivity Studies (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: Deutsche Forschungsgemeinschaft (DFG HI 1928/2-1, HI 1928/6-1, 1928/5-1, HI 1928/2-1, SCHW 1768/8-1, HI 1928/14-1, /14-1, HI 1928/5‐1); Hector Stiftung; Hector Stiftung II
Citations: not cited yet (Europe PMC); 80 references in the paper

Abstract

Brain-age models use neuroimaging features to predict chronological age. The resulting brain-age gap (BAG), defined as predicted brain age minus chronological age, quantifies deviations from age-expected brain characteristics. Structural brain-age acceleration is well established in schizophrenia spectrum disorders (SSD), but the utility of resting-state functional connectivity (rs-FC)–based brain age remains unclear. Here, we trained rs-FC brain-age models on aggregated lifespan data from healthy controls (n ≈ 2200) and evaluated them in four independent SSD case–control cohorts. Across independent cohorts and two brain atlases, SSD showed higher FC-based BAG than healthy controls (β ≈ 0.4–0.6), indicating modest functional brain-age elevation at the group level. However, within SSD, more negative (younger appearing) BAG was associated with poorer cognitive performance, longer duration of illness, and higher neurological soft signs (NSS). Over 12–24 weeks, increases in BAG were associated with reductions in NSS motor coordination and hard signs. Together, these findings suggest that rs-FC brain age captures both a small case–control shift and a clinically relevant dimension within SSD that is not well described by uniform “acceleration”. FC-based BAG may therefore reflect heterogeneity in network-level development and reorganization, with younger-appearing functional profiles potentially indexing greater neurodevelopmental burden.

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

Repository

Its files are read in the Code ↔ Paper reader above, with 1 match between paragraphs and lines of code.

rpomponio/neuroHarmonize

License: MIT
State: the link answers, verified on 30 October 2026
Evidence: files inventoried
Commit: 22c550b68e8413f5a64b279afb7e6796185f6a0d, 6 October 2026
Languages: Python (10)
Size: 24 files, 10 scripts
Software Heritage: archived
Found in: “Code availability”
Holds: README, license file, environment (pyproject.toml, setup.py), tests, continuous integration
Not found: CITATION.cff, documentation
Tools: NumPy (8 files), pandas (7 files), neuroHarmonize (4 files), NiBabel (4 files), neuroCombat (3 files), statsmodels (3 files)
Availability: 2 checks, the latest on 30 October 2026: the link answers
  • 30 October 2026: the link answers
  • 26 September 2026: the link answers
12 files

Code availability

Only publicly available software was used for the preprocessing and analysis of this analysis. Fmriprep was used for preprocessing of the MRI data and is available here https://fmriprep.org/en/23.0.1/, connectivity matrices were calculated with nilearn https://nilearn.github.io/stable/index.html. Data was harmonized with https://github.com/rpomponio/neuroHarmonize. Machine learning age regression was calculated with scikit learn https://scikit-learn.org/stable/index.html. Optuna was used for gridsearch https://hub.optuna.org/samplers/grid_search/ statsmodels were used for statistical testing https://www.statsmodels.org/stable/index.html.

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

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;
  • 10 scripts, each with its path and the digest of its content;
  • 1 match between paragraphs of the paper and lines of the code (method lexical-v1);
  • 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

Datasets cited

Data availability

The inhouse data is not publicly available but can be shared upon reasonable request. The public available datasets can be accessed as follows: CamCAN dataset is available here https://opendata.mrc-cbu.cam.ac.uk/projects/camcan/, AOMIC PIOP1: https://openneuro.org/datasets/ds002785/versions/2.0.0, AOMIC PIOP2: https://openneuro.org/datasets/ds002790/versions/2.0.0, Dallas Lifespan: https://openneuro.org/datasets/ds004856/versions/1.3.0, OASIS-3: https://www.nitrc.org/projects/oasis/, SALD: https://fcon_1000.projects.nitrc.org/indi/retro/sald.html.

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 3, 28 September 2026

  • Funding: added Deutsche Forschungsgemeinschaft: DFG HI 1928/2-1, HI 1928/6-1, 1928/5-1, HI 1928/2-1, SCHW 1768/8-1, HI 1928/14-1, /14-1, HI 1928/5‐1; Hector Stiftung; Hector Stiftung II

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 12 authors, 2 keywords, 14 MeSH terms, 72 references.

Cite

This paper

Volkmer, S., Fritze, S., Altinok, D. C. A., Brandt, G., Kocak, E., Wiegert, J., Berhe, O., Lin, Y., Tost, H., Meyer-Lindenberg, A., Hirjak, D., & Schwarz, E. (2026). Resting-state brain age is associated with cognitive and sensorimotor abnormalities in schizophrenia spectrum disorders: a validation & longitudinal study. Translational psychiatry, 16(1), 451. https://doi.org/10.1038/s41398-026-04426-3

BibTeX

@article{volkmer2026resting,
author = {Volkmer, Sebastian and Fritze, Stefan and Altinok, Dilsa Cemre Akkoc and Brandt, Geva and Kocak, Ersoy and Wiegert, Julius and Berhe, Oksana and Lin, Yuchen and Tost, Heike and Meyer-Lindenberg, Andreas and Hirjak, Dusan and Schwarz, Emanuel},
title = {{Resting-state brain age is associated with cognitive and sensorimotor abnormalities in schizophrenia spectrum disorders: a validation \& longitudinal study}},
journal = {Translational psychiatry},
year = {2026},
month = sep,
volume = {16},
number = {1},
pages = {451},
publisher = {Nature Publishing Group},
issn = {2158-3188},
doi = {10.1038/s41398-026-04426-3},
url = {https://doi.org/10.1038/s41398-026-04426-3},
pmid = {42697895},
pmcid = {PMC13545253}
}

RIS

TY - JOUR
AU - Volkmer, Sebastian
AU - Fritze, Stefan
AU - Altinok, Dilsa Cemre Akkoc
AU - Brandt, Geva
AU - Kocak, Ersoy
AU - Wiegert, Julius
AU - Berhe, Oksana
AU - Lin, Yuchen
AU - Tost, Heike
AU - Meyer-Lindenberg, Andreas
AU - Hirjak, Dusan
AU - Schwarz, Emanuel
TI - Resting-state brain age is associated with cognitive and sensorimotor abnormalities in schizophrenia spectrum disorders: a validation & longitudinal study
T2 - Translational psychiatry
J2 - Transl Psychiatry
PY - 2026
DA - 2026/09/04
VL - 16
IS - 1
SP - 451
SN - 2158-3188
PB - Nature Publishing Group
DO - 10.1038/s41398-026-04426-3
UR - https://doi.org/10.1038/s41398-026-04426-3
LA - en
ER -

CSL-JSON

{
"id": "10.1038/s41398-026-04426-3",
"type": "article-journal",
"title": "Resting-state brain age is associated with cognitive and sensorimotor abnormalities in schizophrenia spectrum disorders: a validation & longitudinal study",
"container-title": "Translational psychiatry",
"author": [
{
"family": "Volkmer",
"given": "Sebastian"
},
{
"family": "Fritze",
"given": "Stefan"
},
{
"family": "Altinok",
"given": "Dilsa Cemre Akkoc"
},
{
"family": "Brandt",
"given": "Geva"
},
{
"family": "Kocak",
"given": "Ersoy"
},
{
"family": "Wiegert",
"given": "Julius"
},
{
"family": "Berhe",
"given": "Oksana"
},
{
"family": "Lin",
"given": "Yuchen"
},
{
"family": "Tost",
"given": "Heike"
},
{
"family": "Meyer-Lindenberg",
"given": "Andreas"
},
{
"family": "Hirjak",
"given": "Dusan"
},
{
"family": "Schwarz",
"given": "Emanuel"
}
],
"container-title-short": "Transl Psychiatry",
"volume": "16",
"issue": "1",
"page": "451",
"DOI": "10.1038/s41398-026-04426-3",
"PMID": "42697895",
"PMCID": "PMC13545253",
"ISSN": "2158-3188",
"publisher": "Nature Publishing Group",
"URL": "https://doi.org/10.1038/s41398-026-04426-3",
"language": "en",
"issued": {
"date-parts": [
[
2026,
9,
4
]
]
}
}

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:10.1038/s41398-026-03956-0 [code]
Normative age-related structural brain deviations underlying psychopathology, cognitive impairment and neurological soft signs in schizophrenia spectrum disorders.
Journal: Translational psychiatry
In common: NiBabel, pandas, NumPy, schizophrenia / psychosis, structural MRI / diffusion, 9 references, 5 authors
[2] doi:10.1038/s41380-026-03691-4 [code]
Breaking the norm: population-scale deviations of brain structure in depression and anxiety.
Journal: Molecular psychiatry
In common: statsmodels, NiBabel, pandas, 1 other tool, structural MRI / diffusion, 1 reference, 3 authors
[3] doi:10.1101/2025.09.26.25336528 [code]
Breaking the Norm: Population-Scale Deviations of Brain Structure in Depression and Anxiety
Journal: medRxiv (preprint)
In common: statsmodels, NiBabel, pandas, 1 other tool, 1 reference, 3 authors
[4] doi:10.3390/jimaging11120454
A Structured Review and Quantitative Profiling of Public Brain MRI Datasets for Foundation Model Development.
Journal: Journal of imaging
In common: OpenNeuro ds002785, OpenNeuro ds002790, 1 other dataset, structural MRI / diffusion, 3 references
[5] doi:10.1016/j.nicl.2025.103881 [code]
Brain age gap reduction following exercise mirrors clinical improvements in schizophrenia spectrum disorders.
Journal: NeuroImage. Clinical
In common: statsmodels, NiBabel, pandas, 1 other tool, schizophrenia / psychosis, structural MRI / diffusion, 7 references
[6] doi:10.1038/s41467-026-72875-x [code]
Lifespan normative modeling of brain microstructure.
Journal: Nature communications
In common: pandas, NumPy, OpenNeuro ds002785, OpenNeuro ds002790, schizophrenia / psychosis, structural MRI / diffusion, 3 references
[7] doi:10.1038/s41586-026-10454-2 [code]
White matter micro- and macrostructure brain charts for the human lifespan.
Journal: Nature
In common: pandas, NumPy, OpenNeuro ds002785, OpenNeuro ds002790, 1 other dataset
[8] doi:10.1038/s42003-025-08231-4 [code]
Structural-functional brain network coupling during cognitive demand reveals intelligence-relevant communication strategies.
Journal: Communications biology
In common: NiBabel, pandas, NumPy, OpenNeuro ds002785, OpenNeuro ds002790, structural MRI / diffusion, 2 references
[9] doi:10.1002/hbm.70519 [code]
Lifespan Trajectories of Asymmetry in White Matter Tracts.
Journal: Human brain mapping
In common: OpenNeuro ds002785, OpenNeuro ds002790, 1 other dataset, structural MRI / diffusion
[10] doi:10.1101/2025.09.29.678806 [code]
Lifespan Trajectories of Asymmetry in White Matter Tracts
Journal: bioRxiv (preprint)
In common: OpenNeuro ds002785, OpenNeuro ds002790, 1 other dataset, structural MRI / diffusion

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.

Add the badge to the code's README

OSCR: the badge of this paper

If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.

Markdown
<!-- oscr-badge:start -->
[![OSCR](https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3.svg)](https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3)
<!-- oscr-badge:end -->
reStructuredText
.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3.svg
   :target: https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3
   :alt: OSCR

.. oscr-badge:end
HTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3"><img src="https://openscicode.org/badge/doi/10.1038/s41398-026-04426-3.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->

On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".

If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge 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: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Report a mismatch between the code and the paper · Report a code error (signed in)

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Report a reproduction failure (signed in)

Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.

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