{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T06:27:11Z","timestamp":1787120831604,"version":"build-2736575974"},"reference-count":43,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T00:00:00Z","timestamp":1787097600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"\n Objective<\/jats:title>\n To develop a multimodal predictive framework that integrates breast magnetic resonance imaging (MRI) and structured clinical features for predicting residual cancer burden (RCB) following neoadjuvant therapy.<\/jats:p>\n <\/jats:sec>\n \n Methods<\/jats:title>\n \n The proposed Breast Cancer Multi-source Multi-scale Model (BCMM) was evaluated on the I-SPY1 cohort (\n n<\/jats:italic>\n = 201), incorporating eight structured clinical variables and 2.5D MRI inputs. The model consists of a clinical feature encoding branch (MLP with squeeze-and-excitation), a Vision Transformer (ViT)-based imaging encoder with multi-scale feature aggregation, and a gated bidirectional cross-attention fusion module. Model performance was assessed using stratified five-fold group cross-validation at the patient level. Additionally, bootstrap resampling was applied to pooled out-of-fold (OOF) predictions to estimate 95% confidence intervals.\n <\/jats:p>\n <\/jats:sec>\n \n Results<\/jats:title>\n Across the five folds, BCMM achieved a mean AUC of 0.826 \u00b1 0.076 and a mean AUPRC of 0.926 \u00b1 0.028. On pooled out-of-fold predictions, BCMM achieved an AUC of 0.795, an AUPRC of 0.895, an accuracy of 0.677, a balanced accuracy of 0.727, a macro-F1 score of 0.662, and an MCC of 0.406.<\/jats:p>\n <\/jats:sec>\n \n Conclusion<\/jats:title>\n BCMM showed promising internal cross-validated discrimination in this single-cohort study. Independent external validation is required before conclusions regarding robustness or clinical utility can be made.<\/jats:p>\n <\/jats:sec>","DOI":"10.3389\/frai.2026.1846986","type":"journal-article","created":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T05:44:27Z","timestamp":1787118267000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["BCMM: gated bidirectional cross-attention fusion of breast MRI and clinical features for predicting RCB response in the I-SPY1 cohort"],"prefix":"10.3389","volume":"9","author":[{"given":"Mu","family":"Yang","sequence":"first","affiliation":[{"name":"School of Medical Information and Engineering, Xuzhou Medical University","place":["Xuzhou, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Junfeng","family":"Hu","sequence":"additional","affiliation":[{"name":"School of Medical Information and Engineering, Xuzhou Medical University","place":["Xuzhou, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,19]]},"reference":[{"key":"B1","unstructured":"Gated multimodal units for information fusion\n \n \n \n Arevalo\n J.\n \n \n Solorio\n T.\n \n \n Montes-y-G\u00f3mez\n M.\n \n \n Gonz\u00e1lez\n F. A.\n \n \n arXiv\n \n 2017"},{"key":"B2","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1175\/1520-0493(1950)078andlt;0001:VOFEITandgt;2.0.CO;2","article-title":"Verification of forecasts expressed in terms of probability","volume":"78","author":"Brier","year":"1950","journal-title":"Mon. Weather Rev."},{"key":"B3","doi-asserted-by":"publisher","first-page":"6","DOI":"10.1186\/s12864-019-6413-7","article-title":"The advantages of the matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation","volume":"21","author":"Chicco","year":"2020","journal-title":"BMC Genomics"},{"key":"B4","doi-asserted-by":"publisher","first-page":"e078378","DOI":"10.1136\/bmj-2023-078378","article-title":"TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods","volume":"385","author":"Collins","year":"2024","journal-title":"BMJ"},{"key":"B5","doi-asserted-by":"publisher","first-page":"113868","DOI":"10.1016\/j.engappai.2026.113868","article-title":"Multimodal exponential moving average-guided representation for integrated tumor-segmentation and survival prediction","volume":"167","author":"Datta","year":"2026","journal-title":"Eng. Appl. Artif. Intell"},{"key":"B6","doi-asserted-by":"publisher","first-page":"233","DOI":"10.1145\/1143844.1143874","article-title":"\u201cThe relationship between precision-recall and ROC curves,\u201d","author":"Davis","year":"2006","journal-title":"Proceedings of the 23rd international conference on machine learning"},{"key":"B7","doi-asserted-by":"publisher","first-page":"837","DOI":"10.2307\/2531595","article-title":"Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach","volume":"44","author":"DeLong","year":"1988","journal-title":"Biometrics"},{"key":"B8","unstructured":"An image is worth 16x16 words: transformers for image recognition at scale\n \n \n \n Dosovitskiy\n A.\n \n \n Beyer\n L.\n \n \n Kolesnikov\n A.\n \n \n Weissenborn\n D.\n \n \n Zhai\n X.\n \n \n Unterthiner\n T.\n \n \n arXiv\n \n 2021"},{"key":"B9","doi-asserted-by":"publisher","first-page":"624","DOI":"10.19300\/j.2025.L22567","article-title":"Early prediction of neoadjuvant chemotherapy efficacy in breast cancer using multi-timepoint ultrafast dynamic contrast-enhanced MRI","volume":"48","author":"Du","year":"2025","journal-title":"Int. J. Med. Radiol"},{"key":"B10","doi-asserted-by":"crossref","DOI":"10.1201\/9780429246593","volume-title":"An Introduction to the Bootstrap.","author":"Efron","year":"1994"},{"key":"B11","doi-asserted-by":"publisher","first-page":"3242","DOI":"10.1200\/JCO.2011.39.2779","article-title":"Pathologic complete response predicts recurrence-free survival more effectively by cancer subset: results from the I-SPY 1 TRIAL\u2013CALGB 150007\/150012, ACRIN 6657","volume":"30","author":"Esserman","year":"2012","journal-title":"J. Clin. Oncol"},{"key":"B12","doi-asserted-by":"publisher","first-page":"458","DOI":"10.1002\/bimj.200410135","article-title":"Estimation of the youden index and its associated cutoff point","volume":"47","author":"Fluss","year":"2005","journal-title":"Biomed. J"},{"key":"B13","unstructured":"On calibration of modern neural networks\n \n \n \n Guo\n C.\n \n \n Pleiss\n G.\n \n \n Sun\n Y.\n \n \n Weinberger\n K. Q.\n \n \n arXiv\n \n 2017"},{"key":"B14","doi-asserted-by":"publisher","first-page":"7132","DOI":"10.1109\/CVPR.2018.00745","article-title":"\u201cSqueeze-and-excitation networks,\u201d","author":"Hu","year":"2018","journal-title":"Proceedings of the IEEE conference on computer vision and pattern recognition"},{"key":"B15","doi-asserted-by":"publisher","first-page":"44","DOI":"10.1148\/radiol.2015150013","article-title":"Neoadjuvant chemotherapy for breast cancer: functional tumor volume by MR imaging predicts recurrence-free survival\u2014results from the ACRIN 6657\/CALGB 150007 I-SPY 1 TRIAL","volume":"279","author":"Hylton","year":"2016","journal-title":"Radiology"},{"key":"B16","doi-asserted-by":"publisher","first-page":"476","DOI":"10.1002\/jmri.24351","article-title":"Optimized breast MRI functional tumor volume as a biomarker of recurrence-free survival following neoadjuvant chemotherapy","volume":"40","author":"Jafri","year":"2014","journal-title":"J. Magn. Reson. Imaging"},{"key":"B17","doi-asserted-by":"publisher","first-page":"18800","DOI":"10.1038\/s41598-021-98408-8","article-title":"Multimodal deep learning models for the prediction of pathologic response to neoadjuvant chemotherapy in breast cancer","volume":"11","author":"Joo","year":"2021","journal-title":"Sci. Rep"},{"key":"B18","doi-asserted-by":"publisher","first-page":"2784","DOI":"10.3390\/tomography8060232","article-title":"Deep learning prediction of pathologic complete response in breast cancer using MRI and other clinical data: a systematic review","volume":"8","author":"Khan","year":"2022","journal-title":"Tomography"},{"key":"B19","doi-asserted-by":"publisher","first-page":"72","DOI":"10.1186\/s13062-025-00661-8","article-title":"Multimodal deep learning for predicting neoadjuvant treatment outcomes in breast cancer: a systematic review","volume":"20","author":"Krasniqi","year":"2025","journal-title":"Biol. Direct"},{"key":"B20","doi-asserted-by":"publisher","first-page":"S62","DOI":"10.1016\/j.acra.2023.02.024","article-title":"Use of pretreatment multiparametric MRI to predict tumor regression pattern to neoadjuvant chemotherapy in breast cancer","volume":"30","author":"Liu","year":"2023","journal-title":"Acad. Radiol"},{"key":"B21","doi-asserted-by":"publisher","first-page":"391","DOI":"10.12435\/j.issn.2095-5227.24120902","article-title":"Correlation factors and multimodal imaging model prediction of neoadjuvant treatment responsiveness in breast cancer","volume":"46","author":"Liu","year":"2025","journal-title":"Acad. J. Chin. PLA Med. Sch"},{"key":"B22","unstructured":"ViLBERT: pretraining task-agnostic visiolinguistic representations for vision-and-language tasks\n \n \n \n Lu\n J.\n \n \n Batra\n D.\n \n \n Parikh\n D.\n \n \n Lee\n S.\n \n \n arXiv\n \n 2019"},{"key":"B23","doi-asserted-by":"publisher","first-page":"153","DOI":"10.1007\/BF02295996","article-title":"Note on the sampling error of the difference between correlated proportions or percentages","volume":"12","author":"McNemar","year":"1947","journal-title":"Psychometrika"},{"key":"B24","doi-asserted-by":"publisher","first-page":"e200029","DOI":"10.1148\/ryai.2020200029","article-title":"Checklist for artificial intelligence in medical imaging (CLAIM): a guide for authors and reviewers","volume":"2","author":"Mongan","year":"2020","journal-title":"Radiol. Artif. Intell"},{"key":"B25","unstructured":"Cross-attention multimodal fusion of mammography and categorical clinical variables for breast cancer classification\n \n \n \n Nantogmah\n M. T.\n \n \n Alhassan\n A.-B.\n \n \n Alhassan\n S.\n \n \n arXiv\n \n 2025"},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.7937\/K9\/TCIA.2016.HdHpgJLK","article-title":"Multi-center breast DCE-MRI data and segmentations from patients in the I-SPY 1\/ACRIN 6657 trials","author":"Newitt","year":"2016","journal-title":"Cancer Imaging Arch"},{"key":"B27","first-page":"689","article-title":"\u201cMultimodal deep learning,\u201d","volume-title":"Proceedings of the 28th international conference on machine learning","author":"Ngiam","year":"2011"},{"key":"B28","doi-asserted-by":"publisher","first-page":"177","DOI":"10.19300\/j.2019.Z6391","article-title":"Research progress of MRI in evaluating the efficacy of neoadjuvant chemotherapy for breast cancer","volume":"42","author":"Peng","year":"2019","journal-title":"Int. J. Med. Radiol."},{"key":"B29","doi-asserted-by":"publisher","first-page":"846775","DOI":"10.3389\/fonc.2022.846775","article-title":"Pretreatment DCE-MRI-based deep learning outperforms radiomics analysis in predicting pathologic complete response to neoadjuvant chemotherapy in breast cancer","volume":"12","author":"Peng","year":"2022","journal-title":"Front. Oncol"},{"key":"B30","doi-asserted-by":"publisher","first-page":"170124","DOI":"10.1038\/sdata.2017.124","article-title":"The public cancer radiology imaging collections of the cancer imaging archive","volume":"4","author":"Prior","year":"2017","journal-title":"Sci. Data"},{"key":"B31","doi-asserted-by":"publisher","first-page":"419","DOI":"10.1002\/bimj.200710415","article-title":"Youden index and optimal cut-point estimated from observations affected by a lower limit of detection","volume":"50","author":"Ruopp","year":"2008","journal-title":"Biomed. J"},{"key":"B32","doi-asserted-by":"publisher","first-page":"e0118432","DOI":"10.1371\/journal.pone.0118432","article-title":"The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets","volume":"10","author":"Saito","year":"2015","journal-title":"PLoS ONE"},{"key":"B33","doi-asserted-by":"publisher","first-page":"102802","DOI":"10.1016\/j.media.2023.102802","article-title":"Transformers in medical imaging: a survey","volume":"88","author":"Shamshad","year":"2023","journal-title":"Med. Image Anal"},{"key":"B34","unstructured":"A disciplined approach to neural network hyper-parameters: part 1 \u2013 learning rate, batch size, momentum, and weight decay\n \n \n \n Smith\n L. N.\n \n \n arXiv\n \n 2018"},{"key":"B35","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-030-16399-0","volume-title":"Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating.","author":"Steyerberg","year":"2019"},{"key":"B36","doi-asserted-by":"publisher","first-page":"4414","DOI":"10.1200\/JCO.2007.10.6823","article-title":"Measurement of residual breast cancer burden to predict survival after neoadjuvant chemotherapy","volume":"25","author":"Symmans","year":"2007","journal-title":"J. Clin. Oncol"},{"key":"B37","first-page":"2818","article-title":"\u201cRethinking the inception architecture for computer vision,\u201d","author":"Szegedy","year":"2015"},{"key":"B38","unstructured":"Attention is all you need\n \n \n \n Vaswani\n A.\n \n \n Shazeer\n N.\n \n \n Parmar\n N.\n \n \n Uszkoreit\n J.\n \n \n Jones\n L.\n \n \n Gomez\n A. N.\n \n \n arXiv\n \n 2023"},{"key":"B39","doi-asserted-by":"publisher","first-page":"565","DOI":"10.1177\/0272989X06295361","article-title":"Decision curve analysis: a novel method for evaluating prediction models","volume":"26","author":"Vickers","year":"2006","journal-title":"Med. Decis. Mak"},{"key":"B40","doi-asserted-by":"publisher","first-page":"1439","DOI":"10.12122\/j.issn.1674-4500.2025.11.19","article-title":"Progress of MRI-based artificial intelligence technology in evaluating the response to neoadjuvant chemotherapy for breast cancer","volume":"48","author":"Wang","year":"2025","journal-title":"J. Mol. Imaging"},{"key":"B41","doi-asserted-by":"publisher","first-page":"694","DOI":"10.1145\/775047.775151","article-title":"\u201cTransforming classifier scores into accurate multiclass probability estimates,\u201d","author":"Zadrozny","year":"2002","journal-title":"Proceedings of the eighth ACM SIGKDD international conference on knowledge discovery and data mining"},{"key":"B42","unstructured":"Mixup: beyond empirical risk minimization\n \n \n \n Zhang\n H.\n \n \n Cisse\n M.\n \n \n Dauphin\n Y. N.\n \n \n Lopez-Paz\n D.\n \n \n arXiv\n \n 2018"},{"key":"B43","doi-asserted-by":"publisher","first-page":"1747","DOI":"10.1186\/s12885-025-15095-8","article-title":"Predicting breast cancer response to neoadjuvant therapy by integrating radiomic and deep-learning features from early-and-peak phases of DCE-MRI","volume":"25","author":"Zhang","year":"2025","journal-title":"BMC Cancer"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1846986\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T05:44:34Z","timestamp":1787118274000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1846986\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,19]]},"references-count":43,"alternative-id":["10.3389\/frai.2026.1846986"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1846986","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,19]]},"article-number":"1846986"}}