Table 1.
Summary of AI methods in breast cancer immunotherapy prediction
| Omic | Biomarker | Task | Data source | Tumor | Model/Algorithm | Year | Reference |
|---|---|---|---|---|---|---|---|
| Transcriptomic | Expression of M2-like tumour-associated macrophages | Immunotherapy response | TCGA | TNBC | Simple neural network-based deep learning | 2020 | Bao et al. 2020) |
| Expression of antigen processing and presentation machinery | Prognostic | TCGA | BC | LASSO logistic regression and RF | 2023 | Müller et al. 2023) | |
| Expression of costimulatory molecule genes | TIME status | TCGA | TNBC | LASSO and SVM-RFE | 2024 | Zhang et al. 2024) | |
| Expression of genes | Immune-related cells | Institutional and public Dataset | BC and melanoma | Gaussian mixture modeling, Kullback–Leibler (KL) divergence, and mutual nearest-neighbors criteria | 2023 | Zou et al. 2023) | |
| Expression of redox-related gene | Prognostic | Institutional and public dataset | BC | RF, LASSO, GBM, Survival-SVM, superpc, Ridge Regression, plsrcox, coxboost, Stepwise Cox regression, and Enet | 2024 | Wang et al. 2024) | |
| Expression of genes | pCR | Institutional and public dataset | BC | LASSO, SVM-RFE, and xgboost | 2024 | Lu et al. 2024) | |
| Expression of DPP4-related genes | Prognosis | Institutional and public dataset | TNBC | Xgboost, RF, adaboost, KNN, ANN, and LR | 2023 | Kang et al. 2023) | |
| Proteomic | IHC and H&E TMA | PD-L1 expression | Public dataset | Breast cancer | CNN model with residual connections | 2022 | Shamai et al. 2022) |
| IHC images | PD-L1 Status | Institutional dataset | Pan-cancer | Weakly supervised deep learning with ResNet-50 and attention-based MIL | 2024 | Ligero et al. 2024) | |
| Expression of signature proteins | Immunotherapy response | Institutional dataset | TNBC | RF | 2024 | Li et al. 2024) | |
| Pathology | H&E WSI | Prognostic | Institutional and TCGA dataset | TNBC | Dual CNN | 2022 | Zhao et al. 2022) |
| WSI | TIL | Clinical studies and TCGA datasets | BC | CNN architectures with segmentation and attention-based classifier | 2024 | Perera et al. 2024) | |
| WSI | Immunotherapy-related gene | TCGA | BC | MIL architecture with resnet-18 | 2025 | Zhang et al. 2025) | |
| H&E TMA | pCR | Clinical studies | TNBC | Regularized logistic regression | 2023 | Wang et al. 2023a) | |
| H&E WSI | lncRNA–metabolism class prediction | Institutional and TCGA dataset | BC | Resnet50 for feature extraction and gated attention | 2024 | Yu et al. 2024) | |
| H&E WSI | TMB | TCGA | TNBC | LR, KNN, RF, and DT | 2025 | Bendani et al. 2025) | |
| H&E WSI | TIL and ki-67 | Public dataset | BC | U-net-like backbone with CNN and residual dilated inception module | 2021 | Negahbani et al. 2021) | |
| Radiomic | DEC-MRI | pCR | Institutional dataset | TNBC | Lasso regression | 2023 | Ramtohul et al. 2024) |
| DCE-MRI | TME subtypes | TCGA | BC | RF | 2024 | Han et al. 2024) | |
| CE-CT | Immunotherapy response | Clinical studies | BC | Multivariable logistic regression | 2023 | Zhao et al. 2023) | |
| MRI | TIL | Institutional dataset | BC | LR, RF, MLP, SVM, LDA, and GB | 2022 | Huang and Lin 2022) | |
| PET/CT | pCR | Institutional dataset | TNBC | LR | 2023 | Seban et al. 2023) | |
| Epigenetic | Methylation level of CpG sites | Prognostic | TCGA | BC | Cox and LASSO regression | 2021 | Zhang et al. 2021) |
| Expression of senescence-relevant lncRNAs | Prognostic | TCGA | BC | Cox and LASSO regression | 2023 | Yu et al. 2023b) | |
| Expression of nine hub lncRNAs | CD8 T-cell levels | TCGA | BC | DT, GBM, GLM, ANN, RF, SVM | 2022 | Chen et al. 2022b) | |
| Expression of Mitochondrial DNA methylation | Prognostic | TCGA | BC | Cox and LASSO regression | 2023 | Ma et al. 2023) | |
| Microbiology | Abundances of microbe | Prognostic | TCGA | BC | Cox and LASSO regression | 2021 | Mao et al. 2022) |
| Genus-level intratumor microbiome | Prognostic | TCGA | BC | Cox and LASSO regression | 2024 | Li et al. 2024) | |
| Omic | 256 multi-modal biomarker signatures from (gene expression, cellular morphometric biomarkers, and abundance of microbe) | Prognostic | TCGA | BC | Multivariate Cox regression | 2022 | Mao et al. 2022) |
| Radiological features and gene expression | Axillary lymph node metastasis | TCGA | BC | Multivariate LR analysis | 2024 | Lai et al. 2024) | |
| Pathology images, lncRNA data, immune cell scores, and clinical features | Prognostic | Institutional and TCGA dataset | BC | CNN architecture with attention-module, concatenation fusion, and a conclusive regressor | 2024 | Yu et al. 2024) | |
| Mammogram, MRI, radiological, histopathological, personal, and clinical data | Pcr | Institutional and public Dataset | BC | Resnet18 architecture for feature extraction, cross-modal knowledge mining, and ensemble model for final prediction | 2024 | Gao et al. 2024) | |
| Genomic | RNA-seq data | Somatic copy number aberrations | TCGA | Pan-cancer | Sequence models and graph neural networks | 2024 | 2024) |
| Whole-exome sequencing and RNA sequencing | Neoantigen immunogenicity | Institutional dataset | Pan-cancer | Ensemble model (LR and xgboost) | 2023 | Jin et al. 2021) | |
| Genomic instability-related lncRNA signature | Prognostic | TCGA | BC | Multivariate Cox regression | 2022 | Jiao et al. 2022) |
TCGA The cancer genome atlas, TNBC Triple-negative breast cancer, BC Breast cancer, LASSO Least absolute shrinkage and selection operator, RF Random forest, SVM-RFE Support vector machine-recursive feature elimination, GBM Gradient boosting machines, SuperPC Supervised principal component, plsRcox partial least squares cox regression,Enet Elastic net, xgBoost extreme gradient boosting, AdaBoost Adaptive boosting, KNN K-nearest neighbor, ANN Artificial neural networks, LR Logistic regression, DT Decision tree, MLP Multilayer perceptron, LDA Linear discriminant analysis, GB Gradient boosting, GLM Generalized linear models, CNN Convolutional neural network, MIL Multiple instance learning, xgBoost extreme gradient boosting, IHC Immunohistochemistry, H&E Hematoxylin and eosin, TMA Tissue microarray, pCR pathological complete response, DPP4 dipeptidyl peptidase 4, TIL Tumor-infiltrating lymphocytes, TME Tumor microenvironment, DCE-MRI Dynamic contrast-enhanced MRI, CT Computed tomography, PET Positron emission tomography, WSI Whole slide imaging, TMB Tumor mutation burden, lncRNA long non-coding RNA