OSCR

Precise disease heterogeneity and progression quantification in MSA and Parkinson's disease using machine learning.

Code ↔ Paper

3 matches 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.

The 3 matches
  1. [1] § Methods › Computation of the heterogeneity (HET) score › Training the ML model ↔ scripts/models.py, lines 318 to 394 · score 0.76 · fold split, AutoGluon, balanced accuracy, best model, winner, F1 score · Report a mismatch
  2. [2] § Results › Performance of ML models and ML-derived heterogeneity (HET) scores ↔ scripts/models.py, lines 200 to 251 · score 0.74 · cross validation, AutoGluon, balanced accuracy, best model, F1 score, fold · Report a mismatch
  3. [3] § Methods › Computation of the heterogeneity (HET) score › Training the ML model ↔ scripts/models.py, lines 35 to 79 · score 0.72 · CatBoost, XGBoost, Random forest, IDs, split, models · 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 · 394 lines · 14 KB · MIT · 3 matches

  1. import os
  2. import shutil
  3. import random
  4. import secrets
  6. import joblib
  7. import matplotlib
  8. import numpy as np
  9. import pandas as pd
  11. from xgboost import XGBClassifier
  12. from lightgbm import LGBMClassifier
  13. from catboost import CatBoostClassifier
  14. from sklearn.ensemble import RandomForestClassifier
  15. from sklearn.model_selection import (
  16. StratifiedGroupKFold,
  17. RandomizedSearchCV,
  18. cross_validate,
  19. )
  20. from sklearn.metrics import accuracy_score, balanced_accuracy_score, f1_score
  21. from autogluon.tabular import TabularPredictor
  22. from tqdm import tqdm
  24. random.seed(42)
  25. np.random.seed(42)
  27. def save_models(models, in_path, prefix):
  28. """Utility to save a dict of models under `<in_path>/CV trained models`."""
  29. path = os.path.join(in_path, "CV trained models")
  30. os.makedirs(path, exist_ok=True)
  31. for name, model in models.items():
  32. joblib.dump(model, os.path.join(path, f"{prefix}_{name}.joblib"))
  35. def run_models(df, target_clf, features, ids_column_name, run_dir):
  36. random.seed(42)
  37. np.random.seed(42)
  39. target_Y_clf = df[target_clf].values
  40. predictors_X_clf = df[features].values
  41. groupings = df[ids_column_name].values # subject-wise grouping
  43. xgb_params = {
  44. "n_estimators": [200, 400],
  45. "max_depth": [3, 10, 50],
  46. "learning_rate": [0.001, 0.01, 0.05],
  47. "subsample": [0.5, 0.7, 0.9, 1.0],
  48. "colsample_bytree": [0.5, 0.7, 0.9, 1.0],
  49. "alpha": [0, 0.1, 0.5, 1, 5, 10],
  50. "lambda": [0, 0.1, 0.5, 1, 5, 10, 20],
  51. }
  52. rf_clf_params = {
  53. "n_estimators": [200, 500],
  54. "max_depth": [None, 50, 100, 200],
  55. "min_samples_leaf": [1, 2, 5, 10, 50, 100],
  56. "min_samples_split": [2, 5, 10, 20, 50, 100],
  57. }
  58. lgbm_params = {
  59. "n_estimators": [200, 400],
  60. "max_depth": [3, 6],
  61. "learning_rate": [0.01, 0.05],
  62. "num_leaves": [50, 100],
  63. "min_child_samples": [5, 10],
  64. "subsample": [0.7, 0.9],
  65. "colsample_bytree": [0.7, 0.9],
  66. }
  67. catboost_params = {
  68. "iterations": [200, 400],
  69. "learning_rate": [0.01, 0.05, 0.1],
  70. "depth": [3, 7, 10],
  71. "l2_leaf_reg": [3, 5, 10],
  72. }
  74. classifiers = {
  75. "XGBoost": (XGBClassifier(random_state=42), xgb_params),
  76. "RandomForest": (RandomForestClassifier(random_state=42), rf_clf_params),
  77. "LGBM": (LGBMClassifier(random_state=42, verbose=-1), lgbm_params),
  78. "CatBoost": (CatBoostClassifier(random_state=42, verbose=0), catboost_params),
  79. }
  81. # 10 different seeds for robustness
  82. seed_list = [secrets.randbits(32) for _ in range(10)]
  84. n_iter_indi = 10
  85. cv_splits_n = 3
  86. all_seeds_summary = []
  88. for rseed in seed_list:
  89. print(f"\n==== Seed {rseed} ====")
  90. seed_dir = os.path.join(run_dir, f"seed_{rseed}")
  91. os.makedirs(seed_dir, exist_ok=True)
  93. sgkf = StratifiedGroupKFold(
  94. n_splits=cv_splits_n,
  95. shuffle=True,
  96. random_state=rseed,
  97. )
  98. cv_splits = list(sgkf.split(predictors_X_clf, target_Y_clf, groups=groupings))
  100. final_results = {}
  101. per_fold_rows = []
  103. # ---------------- AutoGluon (per-fold) ----------------
  104. print("Training AutoGluon with StratifiedGroupKFold (subject-wise)")
  105. f1_scores_ag, acc_scores_ag, bacc_scores_ag = [], [], []
  106. ag_fold_paths, ag_fold_scores = [], []
  108. # safe CPU count for AutoGluon
  109. n_cpus = os.cpu_count() or 1
  110. ag_num_cpus = max(1, n_cpus // 2)
  112. for fold_id, (tr_idx, te_idx) in enumerate(
  113. tqdm(cv_splits, desc=f"AG SGKF seed={rseed}"), 1
  114. ):
  115. fold_dir = os.path.join(seed_dir, f"fold_{fold_id}")
  116. os.makedirs(fold_dir, exist_ok=True)
  118. # save fold splits
  119. df.iloc[tr_idx].to_csv(os.path.join(fold_dir, "train.csv"), index=False)
  120. df.iloc[te_idx].to_csv(os.path.join(fold_dir, "test.csv"), index=False)
  122. X_train, X_test = predictors_X_clf[tr_idx], predictors_X_clf[te_idx]
  123. y_train, y_test = target_Y_clf[tr_idx], target_Y_clf[te_idx]
  125. ag_train = pd.DataFrame(X_train, columns=features)
  126. ag_train[target_clf] = y_train
  128. problem_type = (
  129. "binary" if np.unique(y_train).size == 2 else "multiclass"
  130. )
  131. ag_path = os.path.join(fold_dir, "autogluon")
  133. ag_predictor = (
  134. TabularPredictor(
  135. label=target_clf,
  136. path=ag_path,
  137. problem_type=problem_type,
  138. eval_metric="f1",
  139. verbosity=0,
  140. )
  141. .fit(
  142. train_data=ag_train,
  143. presets="medium_quality",
  144. ag_args_fit={"num_cpus": ag_num_cpus},
  145. )
  146. )
  148. # test metrics
  149. X_test_df = pd.DataFrame(X_test, columns=features)
  150. pred_te = ag_predictor.predict(X_test_df)
  151. f1_te = f1_score(y_test, pred_te, average="weighted")
  152. acc_te = accuracy_score(y_test, pred_te)
  153. bacc_te = balanced_accuracy_score(y_test, pred_te)
  154. f1_scores_ag.append(f1_te)
  155. acc_scores_ag.append(acc_te)
  156. bacc_scores_ag.append(bacc_te)
  158. # train metrics
  159. X_train_df = pd.DataFrame(X_train, columns=features)
  160. pred_tr = ag_predictor.predict(X_train_df)
  161. f1_tr = f1_score(y_train, pred_tr, average="weighted")
  162. acc_tr = accuracy_score(y_train, pred_tr)
  163. bacc_tr = balanced_accuracy_score(y_train, pred_tr)
  165. pd.DataFrame(
  166. [
  167. {
  168. "fold": fold_id,
  169. "f1_weighted": f1_te,
  170. "accuracy": acc_te,
  171. "balanced_accuracy": bacc_te,
  172. }
  173. ]
  174. ).to_csv(
  175. os.path.join(fold_dir, "AutoGluon_metrics.csv"),
  176. index=False,
  177. )
  179. per_fold_rows += [
  180. {
  181. "model": "AutoGluon",
  182. "fold": fold_id,
  183. "split": "train",
  184. "f1_weighted": f1_tr,
  185. "accuracy": acc_tr,
  186. "balanced_accuracy": bacc_tr,
  187. },
  188. {
  189. "model": "AutoGluon",
  190. "fold": fold_id,
  191. "split": "test",
  192. "f1_weighted": f1_te,
  193. "accuracy": acc_te,
  194. "balanced_accuracy": bacc_te,
  195. },
  196. ]
  197. ag_fold_paths.append(ag_path)
  198. ag_fold_scores.append((f1_te, bacc_te))
  200. # pick best AG fold by F1 then balanced accuracy
  201. if ag_fold_scores:
  202. f1_arr = np.array([s[0] for s in ag_fold_scores])
  203. bacc_arr = np.array([s[1] for s in ag_fold_scores])
  204. cand = np.where(f1_arr == f1_arr.max())[0]
  205. best_idx = int(cand[np.argmax(bacc_arr[cand])])
  206. best_ag_path = ag_fold_paths[best_idx]
  207. best_ag_model = TabularPredictor.load(best_ag_path)
  208. else:
  209. best_ag_model = None
  211. final_results["AutoGluon"] = {
  212. "best_model": best_ag_model,
  213. "f1_score": float(np.mean(f1_scores_ag)) if f1_scores_ag else np.nan,
  214. "accuracy": float(np.mean(acc_scores_ag)) if acc_scores_ag else np.nan,
  215. "balanced_accuracy": float(np.mean(bacc_scores_ag)) if bacc_scores_ag else np.nan,
  216. "f1_score_std": float(np.std(f1_scores_ag)) if f1_scores_ag else np.nan,
  217. }
  219. # ---------------- sklearn classifiers (per-fold) ----------------
  220. print("Training classifiers with StratifiedGroupKFold (subject-wise)")
  221. for name, (classifier, param_grid) in tqdm(
  222. classifiers.items(), desc=f"GridSearch seed={rseed}"
  223. ):
  224. search = RandomizedSearchCV(
  225. estimator=classifier,
  226. param_distributions=param_grid,
  227. cv=cv_splits,
  228. n_iter=n_iter_indi,
  229. n_jobs=-1,
  230. verbose=1,
  231. random_state=rseed,
  232. scoring="f1_weighted",
  233. refit=True,
  234. )
  235. search.fit(predictors_X_clf, target_Y_clf)
  237. best_model = search.best_estimator_
  239. scores = cross_validate(
  240. best_model,
  241. predictors_X_clf,
  242. target_Y_clf,
  243. cv=cv_splits,
  244. scoring={
  245. "f1": "f1_weighted",
  246. "accuracy": "accuracy",
  247. "balanced_accuracy": "balanced_accuracy",
  248. },
  249. n_jobs=-1,
  250. return_train_score=True,
  251. )
  253. # write per-fold CSVs
  254. for fold_id, (
  255. f1_tr,
  256. acc_tr,
  257. bacc_tr,
  258. f1_te,
  259. acc_te,
  260. bacc_te,
  261. ) in enumerate(
  262. zip(
  263. scores["train_f1"],
  264. scores["train_accuracy"],
  265. scores["train_balanced_accuracy"],
  266. scores["test_f1"],
  267. scores["test_accuracy"],
  268. scores["test_balanced_accuracy"],
  269. ),
  270. 1,
  271. ):
  272. fold_dir = os.path.join(seed_dir, f"fold_{fold_id}")
  273. os.makedirs(fold_dir, exist_ok=True)
  274. out_path = os.path.join(fold_dir, f"{name}_metrics.csv")
  275. row = pd.DataFrame(
  276. [
  277. {
  278. "fold": fold_id,
  279. "model": name,
  280. "f1_weighted": f1_te,
  281. "accuracy": acc_te,
  282. "balanced_accuracy": bacc_te,
  283. }
  284. ]
  285. )
  286. if os.path.exists(out_path):
  287. row.to_csv(out_path, mode="a", header=False, index=False)
  288. else:
  289. row.to_csv(out_path, index=False)
  291. per_fold_rows += [
  292. {
  293. "model": name,
  294. "fold": fold_id,
  295. "split": "train",
  296. "f1_weighted": f1_tr,
  297. "accuracy": acc_tr,
  298. "balanced_accuracy": bacc_tr,
  299. },
  300. {
  301. "model": name,
  302. "fold": fold_id,
  303. "split": "test",
  304. "f1_weighted": f1_te,
  305. "accuracy": acc_te,
  306. "balanced_accuracy": bacc_te,
  307. },
  308. ]
  310. final_results[name] = {
  311. "best_model": best_model,
  312. "f1_score": float(np.mean(scores["test_f1"])),
  313. "accuracy": float(np.mean(scores["test_accuracy"])),
  314. "balanced_accuracy": float(np.mean(scores["test_balanced_accuracy"])),
  315. "f1_score_std": float(np.std(scores["test_f1"])),
  316. }
  318. # save per-fold split metrics for this seed
  319. per_fold_df = (
  320. pd.DataFrame(per_fold_rows)
  321. .sort_values(["model", "fold", "split"])
  322. .reset_index(drop=True)
  323. )
  324. per_fold_df.to_csv(
  325. os.path.join(seed_dir, "per_fold_split_metrics.csv"),
  326. index=False,
  327. )
  329. # winner for this seed: maximize mean F1, then minimize F1 std
  330. best_name = max(
  331. final_results.keys(),
  332. key=lambda k: (final_results[k]["f1_score"], -final_results[k]["f1_score_std"]),
  333. )
  334. best_model_obj = final_results[best_name]["best_model"]
  335. print(f"[Seed {rseed}] Winner: {best_name}")
  337. # save winner into seed
  338. if best_name == "AutoGluon":
  339. if best_model_obj is None or not getattr(best_model_obj, "path", None):
  340. raise RuntimeError(
  341. "AutoGluon chosen as best but best_model/path missing."
  342. )
  343. src_dir = best_model_obj.path
  344. dest_dir = os.path.join(seed_dir, f"{best_name}_best_classifier")
  345. if os.path.exists(dest_dir):
  346. shutil.rmtree(dest_dir)
  347. shutil.copytree(src_dir, dest_dir)
  348. saved_path = dest_dir
  349. else:
  350. model_path = os.path.join(seed_dir, f"{best_name}_best_classifier.joblib")
  351. joblib.dump(best_model_obj, model_path)
  352. saved_path = model_path
  354. # per-seed summary row
  355. all_seeds_summary.append(
  356. {
  357. "seed": rseed,
  358. "winner_model": best_name,
  359. "saved_to": saved_path,
  360. "f1_score": final_results[best_name]["f1_score"],
  361. "f1_score_std": final_results[best_name]["f1_score_std"],
  362. "accuracy": final_results[best_name]["accuracy"],
  363. "balanced_accuracy": final_results[best_name]["balanced_accuracy"],
  364. }
  365. )
  367. winners_df = pd.DataFrame(all_seeds_summary)
  368. winners_df.to_csv(
  369. os.path.join(run_dir, "seeds_winners_summary.csv"),
  370. index=False,
  371. )
  373. # pick overall best across seeds: max F1, then min F1 std
  374. best_row = max(
  375. all_seeds_summary,
  376. key=lambda r: (r["f1_score"], -r["f1_score_std"]),
  377. )
  378. overall_name = best_row["winner_model"]
  379. overall_src = best_row["saved_to"]
  381. if overall_name == "AutoGluon":
  382. dest_dir = os.path.join(run_dir, f"overall_best_{overall_name}_best_classifier")
  383. if os.path.exists(dest_dir):
  384. shutil.rmtree(dest_dir)
  385. shutil.copytree(overall_src, dest_dir)
  386. print("Overall best AutoGluon copied to:", dest_dir)
  387. else:
  388. dest_path = os.path.join(
  389. run_dir,
  390. f"overall_best_{overall_name}_best_classifier.joblib",
  391. )
  392. if os.path.abspath(overall_src) != os.path.abspath(dest_path):
  393. shutil.copy2(overall_src, dest_path)
  394. print("Overall best sklearn model copied to:", dest_path)

models.py at commit 6e4a193, under MIT · at the source

Overview

Authors: Robel K. Gebre1, Sheelakumari Raghavan1, Mari E. Johnson De Tora1, Angela J. Fought2, Robert I. Reid3, Phillip A. Low4, Wolfgang Singer4, Prashanthi Vemuri1
ORCID iDs: Robel K. Gebre
  1. Department of Radiology, Mayo Clinic,Rochester, MN USA
  2. Department of Quantitative Health Sciences, Mayo Clinic,Rochester, MN USA
  3. Department of Information Technology, Mayo Clinic,Rochester, MN USA
  4. Department of Neurology, Mayo Clinic,Rochester, MN USA
Institutions: Mayo Clinic (United States)
Journal: Scientific reports, volume 16, issue 1, article 10579
Dates: received 9 December 2025; accepted 23 March 2026; published online 30 March 2026
Type: Research article · Language: English
License: CC BY-NC-ND
Identifiers: DOI 10.1038/s41598-026-45949-5 · PMID 41912595 · PMCID PMC13039417 · OpenAlex W7143468365
Open access: gold, a free copy (OpenAlex)
Status: code verified (how the code is found and verified)
Categories: structural MRI / diffusion (modality), human (organism), Parkinson's (population), clinical / translational (subfield)
Methods: Statistics, Machine learning, Preprocessing, Connectivity, fMRI & imaging
Keywords: Computational biology and bioinformatics, Diseases, Neurology, Neuroscience
MeSH: Machine Learning*, Multiple System Atrophy*, Parkinson Disease*, Aged, Brain, Diffusion Magnetic Resonance Imaging, Disease Progression, Female, Humans, Magnetic Resonance Imaging, Male, Middle Aged (* major topic)
Topic: Parkinson's Disease Mechanisms and Treatments (Clinical Neurology, Medicine), according to OpenAlex
Funding: National Institute of Neurological Disorders and Stroke (R01NS092625, R01NS097495); Michael J. Fox Foundation for Parkinson’s Research; Bischof Dr. Karl Golser Stiftung
Citations: not cited yet (Europe PMC); 59 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

RobelGebre/HET

License: MIT
State: the link answers, verified on 30 September 2026
Evidence: files inventoried
Commit: 6e4a19345fc7146a9bafeaa8f0920c5407a1b729, 31 March 2026
Languages: Python (3)
Size: 13 files, 3 scripts
Software Heritage: not archived
Found in: the text, “Training the ML model”
Holds: README, environment (requirements.txt)
Not found: license file, CITATION.cff, tests, continuous integration, documentation
Tools: NumPy (3 files), pandas (3 files), scikit-learn (3 files), Matplotlib (2 files), LightGBM (1 file), seaborn (1 file), SHAP (1 file), XGBoost (1 file)
Availability: 1 check, the latest on 30 September 2026: the link answers
  • 30 September 2026: the link answers
4 files
At the source: github.com/RobelGebre/HET

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;
  • 3 scripts, each with its path and the digest of its content;
  • 3 matches 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

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

Code and data availability statement

The paper has a code and data availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

  • it says that the data are available on request

Read it in the paper: doi.org/10.1038/s41598-026-45949-5.

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, 30 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 8 authors, 4 keywords, 12 MeSH terms, 3 funders, 55 references.

Cite

This paper

Gebre, R. K., Raghavan, S., De Tora, M. E. J., Fought, A. J., Reid, R. I., Low, P. A., Singer, W., & Vemuri, P. (2026). Precise disease heterogeneity and progression quantification in MSA and Parkinson's disease using machine learning. Scientific reports, 16(1), 10579. https://doi.org/10.1038/s41598-026-45949-5

BibTeX

@article{gebre2026precise,
author = {Gebre, Robel K. and Raghavan, Sheelakumari and De Tora, Mari E. Johnson and Fought, Angela J. and Reid, Robert I. and Low, Phillip A. and Singer, Wolfgang and Vemuri, Prashanthi},
title = {{Precise disease heterogeneity and progression quantification in MSA and Parkinson's disease using machine learning}},
journal = {Scientific reports},
year = {2026},
month = mar,
volume = {16},
number = {1},
pages = {10579},
publisher = {Nature Publishing Group},
issn = {2045-2322},
doi = {10.1038/s41598-026-45949-5},
url = {https://doi.org/10.1038/s41598-026-45949-5},
pmid = {41912595},
pmcid = {PMC13039417}
}

RIS

TY - JOUR
AU - Gebre, Robel K.
AU - Raghavan, Sheelakumari
AU - De Tora, Mari E. Johnson
AU - Fought, Angela J.
AU - Reid, Robert I.
AU - Low, Phillip A.
AU - Singer, Wolfgang
AU - Vemuri, Prashanthi
TI - Precise disease heterogeneity and progression quantification in MSA and Parkinson's disease using machine learning
T2 - Scientific reports
J2 - Sci Rep
PY - 2026
DA - 2026/03/30
VL - 16
IS - 1
SP - 10579
SN - 2045-2322
PB - Nature Publishing Group
DO - 10.1038/s41598-026-45949-5
UR - https://doi.org/10.1038/s41598-026-45949-5
LA - en
ER -

CSL-JSON

{
"id": "10.1038/s41598-026-45949-5",
"type": "article-journal",
"title": "Precise disease heterogeneity and progression quantification in MSA and Parkinson's disease using machine learning",
"container-title": "Scientific reports",
"author": [
{
"family": "Gebre",
"given": "Robel K."
},
{
"family": "Raghavan",
"given": "Sheelakumari"
},
{
"family": "De Tora",
"given": "Mari E. Johnson"
},
{
"family": "Fought",
"given": "Angela J."
},
{
"family": "Reid",
"given": "Robert I."
},
{
"family": "Low",
"given": "Phillip A."
},
{
"family": "Singer",
"given": "Wolfgang"
},
{
"family": "Vemuri",
"given": "Prashanthi"
}
],
"container-title-short": "Sci Rep",
"volume": "16",
"issue": "1",
"page": "10579",
"DOI": "10.1038/s41598-026-45949-5",
"PMID": "41912595",
"PMCID": "PMC13039417",
"ISSN": "2045-2322",
"publisher": "Nature Publishing Group",
"URL": "https://doi.org/10.1038/s41598-026-45949-5",
"language": "en",
"issued": {
"date-parts": [
[
2026,
3,
30
]
]
}
}

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:10.1093/braincomms/fcae300 [code]
Can integration of Alzheimer's plasma biomarkers with MRI, cardiovascular, genetics, and lifestyle measures improve cognition prediction?
Journal: Brain communications
In common: SHAP, seaborn, scikit-learn, 3 other tools, structural MRI / diffusion, 2 references, author Robel K. Gebre
[2] doi:10.1016/j.isci.2025.114100 [code]
Toward neuroimaging-based diagnostic support: a deep learning approach with a closed-loop system for psychiatric classification.
Journal: iScience
In common: LightGBM, SHAP, XGBoost, 5 other tools, clinical / translational, 2 references
[3] doi:10.7554/elife.81067 [code]
Biological brain age prediction using machine learning on structural neuroimaging data: Multi-cohort validation against biomarkers of Alzheimer's disease and neurodegeneration stratified by sex.
Journal: eLife
In common: LightGBM, SHAP, XGBoost, 4 other tools, structural MRI / diffusion, clinical / translational, 2 references
[4] doi:10.1038/s41597-022-01695-7 [code]
An analysis-ready and quality controlled resource for pediatric brain white-matter research.
Journal: Scientific data
In common: SHAP, XGBoost, seaborn, 4 other tools, structural MRI / diffusion, 3 references
[5] doi:10.1038/s41565-025-01955-8 [code]
Nanoneedles enable spatiotemporal lipidomics of living tissues.
Journal: Nature nanotechnology
In common: LightGBM, SHAP, XGBoost, 5 other tools, clinical / translational, 1 reference
[6] doi:10.3390/s25072332 [code]
Robust EEG Characteristics for Predicting Neurological Recovery from Coma After Cardiac Arrest.
Journal: Sensors (Basel, Switzerland)
In common: LightGBM, SHAP, XGBoost, 5 other tools, clinical / translational, 1 reference
[7] doi:10.1016/j.nicl.2024.103650 [code]
Prognostic enrichment for early-stage Huntington's disease: An explainable machine learning approach for clinical trial.
Journal: NeuroImage. Clinical
In common: LightGBM, SHAP, XGBoost, 5 other tools, clinical / translational, 1 reference
[8] doi:10.1126/sciadv.adu2937 [code]
Gut-brain nexus: Mapping multimodal links to neurodegeneration at biobank scale.
Journal: Science advances
In common: LightGBM, SHAP, XGBoost, 5 other tools, Parkinson's, clinical / translational
[9] doi:10.3389/fphar.2024.1412725 [code]
Neural effects of dopaminergic compounds revealed by multi-site electrophysiology and interpretable machine-learning.
Journal: Frontiers in pharmacology
In common: LightGBM, SHAP, XGBoost, 5 other tools, 1 reference
[10] doi:10.1073/pnas.2516601123 [code]
Unveiling the glymphatic system's role in brain aging: A comprehensive biomarker and modifiable intervention target.
Journal: Proceedings of the National Academy of Sciences of the United States of America
In common: LightGBM, SHAP, XGBoost, 5 other tools, structural MRI / diffusion, clinical / translational

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/s41598-026-45949-5.svg)](https://openscicode.org/badge/doi/10.1038/s41598-026-45949-5)
<!-- oscr-badge:end -->
reStructuredText
.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.1038/s41598-026-45949-5.svg
   :target: https://openscicode.org/badge/doi/10.1038/s41598-026-45949-5
   :alt: OSCR

.. oscr-badge:end
HTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.1038/s41598-026-45949-5"><img src="https://openscicode.org/badge/doi/10.1038/s41598-026-45949-5.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.