This work investigates MRI-based Microbial Density Stratification as a patient-level representation learning task, and Center Heatmap-driven Macro-micro modeling Network (CHM-Net) is introduced for this task, establishing the link between imaging phenotypes and microbial states through center heatmap-guided small-lesion response localization.
Abstract
Microbial density is clinically important for tumor assessment and treatment decision-making, and recent advances in deep learning suggest that it can be non-invasively inferred from multimodal MRI. In this work, MRI-based Microbial Density Stratification (MRI-MDS) is first investigated as a patient-level representation learning task, and Center Heatmap-driven Macro-micro modeling Network (CHM-Net) is introduced for this task. CHM-Net first establishes the link between imaging phenotypes and microbial states through center heatmap-guided small-lesion response localization. Building upon this, it constructs patient-level macro-micro evidence from localized heatmap responses for microbial density prediction. Experiments on the novel GBNPC 2026 dataset constructed for MRI-MDS demonstrate the effectiveness of CHM-Net, achieving superior performance over representative baselines with a 12.06% absolute ACC gain over the strongest competing result. Additionally, auxiliary validation on two 3D medical image datasets further verifies its robustness across volumetric medical image classification scenarios. The project is available at https://anonymous.4open.science/r/CHM-Net-942E/.
Background Cancer recurrence and distant metastasis are major causes of cancer-related death, yet existing biomarkers and single-omics models have limited accuracy and interpretability across tumor types. Methods We developed OMNIS (OMics Network Integration and Spatial representation), a convolutional deep-learning framework that embeds multi-omics profiles into a five-channel genomic image ordered by Hi-C–derived chromosomal proximity. Somatic mutation, copy-number alteration, DNA methylation and gene-expression data from 1,578 TCGA tumors across 33 cancer types were used to train classifiers for recurrence risk and for primary-versus-metastatic status. Performance was assessed by 10-fold cross-validation using AUROC, AUPR and threshold-based metrics. Integrated gradients yielded per-gene attribution scores; top-ranked genes were evaluated for prognostic value in two independent non-small cell lung cancer cohorts (GSE31210, n = 226; GSE135222, n = 27) using survival analyses. Results OMNIS achieved high discrimination for recurrence (AUROC/AUPR 0.970/0.937) and metastasis (0.980/0.883), with accuracies of 0.873–0.911 and negative predictive values ≥0.970 across tasks. Spatial genomic embedding accelerated convergence and outperformed non-spatial baselines. Attribution highlighted seven recurrence-associated genes (including IBA57, DNTTIP1, SLC20A2 and TMEM201) and ten metastasis-associated genes (including PLXNA1, POLR3D, TTLL4, SREBF2, TYMP and ZBTB7C). In external cohorts, expression of these genes showed independent, stage-dependent associations with progression-free and overall survival. Conclusion OMNIS is a spatially informed multi-omics framework that couples accurate prediction with gene-level interpretability. By embedding three-dimensional genome organization into deep-learning models, OMNIS nominates biologically coherent, context-specific drivers of progression and may guide future biomarker development and personalized therapy in precision oncology.
Junxian Li, Yuchen Xing, Ximin Gao et al.· Frontiers in Artificial Inte...· 0 citations
As the volume of omics data continues to grow exponentially, there is an increasing demand for innovative methodologies that combine multi-omics data to extract meaningful clinical insights. Absolute cell counts are a fundamental component of clinical evaluations for disease diagnosis, treatment, and patient management. While cellular deconvolution can estimate relative cell type proportions from bulk data, obtaining absolute cell counts from omics data remains rarely studied. In response to the clinical needs and challenges, we introduce a novel multi-modal deep learning model with intermediate fusion: multi-omics fusion neural network- computational cell counting (MOFUN-CCC). This model is designed to predict absolute cell counts directly by integrating gene expression and DNA methylation data within a supervised framework, assuming that the underlying true cell components are shared across the two omics data. Comprehensive evaluations, including cross-validation, independent data testing, and real-world applications, demonstrate the model's robustness, precision, and capacity to effectively capture biological variations. MOFUN-CCC represents a pioneering effort in the integration of multi-omics data for the prediction of absolute cell counts. With our user-friendly software (https://github.com/yuemolin/MOFUN-CCC) and web application (https://shiny.crc.pitt.edu/mofun_shiny/), this innovation holds the potential to make significant contributions to disease diagnosis, progression analysis, and clinical decision-making.
Predicting cancer drug responses (CDRs) accurately remains a significant challenge due to the complexity of tumor biology and the limitations of existing "black-box" machine learning models. To address this, we propose ProphDR, an interpretable deep learning framework that integrates multiomics data and drug structural information using a hierarchical attention mechanism. ProphDR incorporates a Criss-Cross Gene-level Multiomics Integration (CGMI) module to capture gene-level features and a cross-attention (CA) module to model drug-gene interactions. Evaluated on datasets from GDSC and CCLE, ProphDR achieves state-of-the-art performance in predicting ln(IC50) values (PCC = 0.938, RMSE = 0.978) and classifying drug sensitivity (AUC = 0.981). It also demonstrates strong generalizability in cold-start scenarios involving unseen drugs or cell lines. Crucially, ProphDR generates biologically interpretable attention maps that highlight key pharmacophores and resistance-related genes such as ERBB2 (HER2), consistent with established mechanisms in NSCLC and BRCA. These insights bridge genomic features with phenotypic outcomes, offering valuable guidance for target prioritization and drug repurposing. ProphDR represents a robust and explainable AI tool for advancing precision oncology.
Yundian Zeng, Qing Ye, Jike Wang et al.· Journal of Chemical Informat...· 0 citations
Integrating complex, multi-omics data presents significant challenges. Existing approaches often face a trade-off between model interpretability and representational capacity, with most either relying on post-hoc interpretation or use linear models that may overlook complex interactions. We report Pathway Activity Autoencoders for the multi-omics setting, which embed prior knowledge via pathway-informed architectural constraints, fostering interpretability, while preserving representational power. Our multi-omic framework is applied in the context of breast cancer and is evaluated in survival prediction and subtype classification with results indicating a positive effect of integration. We conduct analysis of individual omics layer impact on end-task performance, revealing that gene, protein, and microRNA expression layers provide the strongest contribution. Repeatability studies indicate that, while dropout improves model robustness and consistency, excessive regularisation can reduce predictive performance. Finally, visualizations of the learned feature space illustrate the framework's intrinsic transparency and clinical relevance. The results underscore the value of multi-omic integration and delineate the impact of individual omics layers, establishing practical guidelines for integration within our framework. Overall, our pathway activity autoencoder frameworks yield superior latent representations that are biologically meaningful and are directly translatable into clinically relevant insights.
Pedro Henrique da Costa Avelar, Ou-Yang Le, Min Wu et al.· 0 citations
Single-cell ribonucleic acid sequencing (scRNA-seq) is a foundational technology for precision-medicine workflows that contribute to United Nations Sustainable Development Goal 3 on Good Health and Well-being, and unsupervised clustering is the analytical step that turns raw expression matrices into interpretable cell populations. Practitioners therefore face a recurring engineering decision: is an additional deep representation stage worth its compute and tuning cost, or do classical principal component analysis (PCA) pipelines already suffice? We address this question with a diagnostic benchmark of nine clustering pipelines on ten real datasets (90-5,685 cells, 19,046-41,480 genes, 4-11 cell types), augmented by a partial scVI V2 specialized comparison on seven datasets. The protocol integrates Optuna hyperparameter search, repeated-run robustness, Friedman/Wilcoxon-Holm/TOST testing, and Sobol total-order sensitivity analysis. The contrastive autoencoder achieved the highest mean Adjusted Rand Index (0.7872), but Holm-corrected tests did not establish dominance over the strongest baselines. Per-dataset analysis reveals three reproducible regimes: probabilistic variational autoencoder (VAE) variants help on the smallest datasets, deep autoencoders win on mid-scale data with multi-batch or many-type structure, and classical PCA pipelines remain competitive when linear projection already captures the dominant variation. Sobol indices identify learning rate ($S_T=0.70$) and latent dimensionality ($S_T=0.56$) as the dominant variance contributors, indicating where limited tuning budgets should be allocated. The contribution is therefore a dataset-aware and compute-conscious decision framework for biomedical AI pipelines supporting sustainable healthcare analytics, rather than a universal superiority claim.
N. Phong, T. Vu, Nguyen Ha Thu et al.· 0 citations
ABSTRACT Background TFE3‐rearranged renal cell carcinoma (TFE3‐rRCC) is a rare, aggressive subtype that predominantly affects adolescents and young adults. Its marked morphologic heterogeneity can delay recognition and downstream confirmatory testing. Methods We assembled a two‐center retrospective cohort of patients < 30 years with renal cell carcinoma (n = 228; 59 TFE3‐rRCC), using fluorescence in situ hybridization (FISH) as the reference standard. Model development was performed in a development cohort (n = 129), followed by independent external validation (n = 99). We developed TFE3‐DualNet, an ensemble of weakly supervised CLAM models trained on routine hematoxylin and eosin (H&E) whole‐slide images (WSIs) using patch embeddings extracted from two pathology foundation models (UNI and CHIEF). We compared performance with three immunohistochemistry (IHC) scoring methods and a feature‐fusion CLAM baseline using concatenated H&E‐derived UNI and CHIEF features, and assessed interpretability by attention mapping. Results In the external validation cohort, TFE3‐DualNet achieved an area under the receiver operating characteristic curve (AUROC) of 0.932, with accuracy 0.879, sensitivity 0.893, and specificity 0.873. The model outperformed IHC scoring methods (AUROC 0.793–0.819; all p < 0.05) and exceeded the feature‐fusion baseline (AUROC 0.906). Attention hotspots localized to diagnostically relevant tumor regions and showed concordance with TFE3 IHC patterns. Conclusions TFE3‐DualNet showed encouraging performance as an interpretable H&E WSI‐based screening model for TFE3‐rRCC in young patients, supporting its potential use to prioritize confirmatory testing and pathologist review in routine diagnostic workflows.
Yu-Hang Chen, Quanhui Xu, Haohua Yao et al.· Cancer Medicine· 0 citations
A new method for surgically removing training examples from a model reveals that as datasets grow, the link between what a model learns and what it produces dissolves.
