SEARCH · PubMed Health

Results for “Deep Learning”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 217 records · Page 12Linked to original sources

PubMed2026

Decoding spatiotemporal fibrotic and cellular immunosuppression of therapeutic T cells in live pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDA) is profoundly immunosuppressive. To help define this behavior, we present integrated experimental and computational frameworks to elucidate therapeutic T cell dynamics. Through the development of TME-CARTographer (TME-CART), a computational pipeline integrating high-dimensional data, graph theory, behavior analysis, and deep learning (DL), we present quantitative insights on 4D T cell-TME interactions in live PDA tumors. Mapping physical immunosuppression demonstrates that collagen fiber architectures direct migration while concomitantly limiting off-axis movement, creating immune exclusion zones. Expanding these findings, we establish that the collagen matrix harbors and spatially organizes immunosuppressive myeloid cells to serve as cooperative co-modulators of T cell behaviors, including migration, sampling, repulsion, and sequestration. Consistent with these findings, DL defines both linear and nonlinear collagen matrix and cellular neighborhood interactions as drivers of T cell behavior. The TME-CART DL framework also accurately predicts shifts in immunosuppression following depletion of myeloid cells. Overall, we identify synergistic barriers impeding anti-tumor T cell behaviors and present TME-CART as a discovery platform for interpreting complex 4D data to enhance the understanding and design of immunotherapies.

Journal Article↗

PubMed2026

Rapidly evolving aphid gall effector proteins exhibit saposin-like folds.

Many insects manipulate plants by injecting effector proteins. In one extreme example of this molecular "hijacking," Hormaphis cornu aphids inject bicycle proteins into Hamamelis virginiana, contributing to the development of novel organs called galls. Bicycle proteins share no amino acid sequence similarity with proteins of known function. Here, we report the crystal structures of two divergent bicycle proteins. Both proteins contain saposin-like folds: one with multiple disulfide bonds exhibits a swapped domain topology; the other has no disulfide bonds and possesses two distinct, tandem domains. To explore the structural evolution of bicycle proteins, we attempted to predict bicycle protein structures with Alphafold2 (AF2) and other deep learning programs. While AF2 did not recover the two experimental structures using existing databases, it succeeded when provided with multiple sequence alignments (MSAs) of protein sequences from newly sequenced closely related species. Using this approach, we generated 2,400 high-confidence bicycle protein predictions from seven aphid species. While all aphid bicycle proteins contain predicted saposin-like folds, they display a vast diversity of structural and physicochemical properties. While this diversity thwarts prediction of conserved functions encoded in structure, it suggests that bicycle proteins have evolved to target diverse plant processes and/or to evade plant immune surveillance. Our extension of AF2 with custom MSAs of proteins from closely related species provides a generalizable, powerful approach for predicting structures of rapidly evolving protein families.

Animals↗

PubMed2005

Influence of community experiences on first-year medical students' reflective writing.

The purpose of this research was to investigate the usefulness of providing students with community-located experiences as a basis for reflection by observing the degree to which community-located experiences influenced the quality of a reflective writing exercise. Students were prepared for reflection by means of a lecture and a handout on the meaning, function and process of reflection. One hundred and twenty-eight students (66 Afrikaans-speaking, and 62 English-speaking) took part in the study. Initially, most students (71%) who revealed affect prior to the visit reported negative emotional states. For 54% of the students who revealed emotions upon arrival, positive emotional states replaced the negative and this increased to 67% as the visit continued. This represented the most important stage of the personal growth continuum, signifying awareness of perspective 'distortions'. The majority (94.3%) regarded the visit as being important prior to immersing themselves in the theory of primary healthcare. In the questionnaire survey, students were asked for their opinions on the value of reflective writing as a tool for promoting deep learning, as students' attitudes were considered an important determinant of increased uptake of this kind of activity. Reflective writing can be promoted through using a real-world experience as stimulus, and a framework for guiding students' thoughts.

Adult↗

PubMed2001

The patient-centred interview: the key to biopsychosocial diagnosis and treatment.

The article reports on some ideas and experiences gained from a holistic approach to working with patients and introduces a viewpoint that includes opinions on how postmodernism, the biopsychosocial model and a patient-centred interviewing style can change traditional, biomedical-oriented medicine. During the past 10 years, we have been instructing medical students in the use of this patient-centred interviewing model and have trained experienced general practitioners (GPs) in adopting it in 2-year family-oriented continuing medical education courses. We believe that doctors and other health care providers, particularly in primary care settings, need a comprehensive concept of human health and illness, and that skill in patient-centred interviewing is the product of a deep learning process. In conclusion, we have learned that a successful patient-centred interview helps the GP to better understand the patient and helps to explain the data that the patient presents. Patient-centred orientation and interviewing also change the communication between doctor and patient in a direction which supports the patient's and his/her family members' own resources in the healing process.

Clinical Competence↗

PubMed2026

Mutational signatures in blood-brain barrier: mechanisms, computational insights, and clinical applications in precision oncology.

The blood - brain barrier (BBB) plays a central role in maintaining central nervous system (CNS) homeostasis, and its disruption is a defining feature of malignant brain tumors such as glioblastoma. Emerging evidence indicates that BBB dysfunction not only alters the tumor microenvironment but also shapes the mutational processes that drive genomic instability in CNS malignancies. This review synthesizes current understanding of the biological mechanisms linking BBB breakdown with distinct mutational signatures, including those arising from oxidative stress, hypoxia-induced replication stress, lipid peroxidation, inflammation, and metabolic reprogramming. Advances in next-generation sequencing, coupled with computational tools such as non-negative matrix factorization, Bayesian modeling, and deep learning, have enabled precise extraction of these signatures and their integration with multi-omics data. Clinically, BBB-associated mutational signatures offer significant promise for therapeutic stratification, prediction of treatment response, and noninvasive monitoring through cerebrospinal fluid - derived circulating tumor DNA. Despite these advances, challenges persist due to limited tissue accessibility, low-yield CSF samples, incomplete mechanistic models, and the lack of CNS-specific analytical frameworks. A deeper understanding of BBB-driven mutational processes, supported by improved computational approaches and integrative datasets, holds potential to advance precision oncology in neuro-oncology.

Humans↗

PubMed2026

A multi-scale fusion model based on multi-phase contrast-enhanced CT for predicting pancreatic cancer resectability.

Purpose.Develop a multi-scale fusion model (MSFM) based on multi-phase contrast-enhanced computed tomography (CECT) to predict pancreatic cancer (PC) resectability, thereby assisting expert decision-making.Methods.This retrospective study enrolled 280 patients with PC from four institutions, which were randomly divided into a training cohort (202 patients) and an independent test cohort (78 patients). Three-phase CECT images (arterial, venous, and delayed phases) were used for modeling. The MSFM comprises two sub-networks: (1) a multi-phase fusion network for extracting cross-phase shared fusion features, (2) a phase-specific branch network for capturing phase-specific features; and a post-fusion strategy to generate the final predictive score by integrating the shared fusion features and three groups of phase-specific features. Additionally, a human-machine fusion deep learning model (HMfDL) was constructed by fusing the predictive score of the MSFM with expert assessments.Results.In the independent test, the MSFM achieved an AUC (area under the receiver operating characteristic curve) of 0.8385 (95% CI: 0.7521-0.9249), accuracy of 84.62%, sensitivity of 72.00%, and specificity of 90.57%. This performance outperformed single-phase models (AUC range: 0.7638-0.7781), two-phase models (AUC range: 0.7826-0.7864), and ten states-of-the-art classifiers (AUC range: 0.7404-0.7796). The HMfDL further improved the performance, reaching an AUC of 0.8626 (95% CI: 0.7853-0.9400), accuracy of 91.03%, sensitivity of 80.00%, and specificity of 96.23%. Notably, the HMfDL corrected 58.82% of misdiagnosis made by experts.Conclusions. The MSFM effectively fuses multi-phase CECT to enable highly accurate predictions of PC resectability, and provides valuable support for expert decision-making through HMfDL.

Humans↗

PubMed2025

Smarter stomata: emergent technologies unlocking yield potential in a changing climate.

Stomata, the gatekeepers of leaf gas exchange, regulate carbon dioxide uptake and water loss, functions increasingly critical as crops face more frequent, intense heat and drought. Under dry conditions, stomatal conductance (g s) typically decreases, limiting carbon assimilation and yield. Heat stress, in contrast, elicits variable g S responses: sometimes increasing to facilitate transpirational cooling, while at other times decreasing, especially when combined with drought. Heat and drought also induce complex, context-dependent shifts in stomatal anatomy. Smaller, denser stomata improve drought resilience in some cases, while reduced density confers greater tolerance in others. The optimal stomatal ideotype remains unknown, and different or even opposing traits may confer resilience dependent on the environmental scenario. Substantial genotypic variation in g s and stomatal anatomy, high heritability and co-localized quantitative trait loci for stomatal traits and yield highlight their untapped potential as breeding targets for climate-resilient crops. However, stomatal traits remain largely absent from breeding pipelines due to challenges of phenotyping at scale. This is changing rapidly. Advances in deep learning, porometry, digital microscopy, and remote sensing now enable high-throughput measurement of stomatal physiology and anatomy. Next-generation breeding technologies including clustered regularly interspaced short palindromic repeats (CRISPR), multi-omics approaches, and artificial intelligence-driven ideotype selection models could revolutionize breeding, allowing precise engineering of stomatal traits for resilience to environmental stress. The time has come to move beyond characterizing stomatal traits and start actively incorporating them into breeding strategies. By leveraging these technologies, stomatal traits can become high value targets, unlocking their potential to enhance crop performance in a hotter, drier future.

abiotic stress↗

PubMed2026

Comprehensive evaluation of AlphaFold/OpenFold prediction of experimentally unresolved proteins through novel metrics.

Predicting accurate protein structures is essential for understanding molecular mechanisms, interpreting the impact of sequence variation, and supporting translational applications ranging from drug discovery to clinical genomics. Recent advances in deep-learning-based predictors such as AlphaFold2, OpenFold, and AlphaFold3 have transformed structural biology, enabling routine in silico modeling even for challenging or previously uncharacterized proteins. However, systematic benchmarking of these tools-especially for novel targets and single amino acid variants-remains limited. Conventional global metrics often fail to capture biologically meaningful discrepancies. By evaluating multiple implementations of AlphaFold2 and OpenFold, together with ColabFold and the AlphaFold3 server, across 10 different proteins and 222 single amino acid protein variants encompassing a wide range of sizes, structures, and functions, we show that although widely used global indicators-like mean pLDDT, pTM-score, and RMSD-frequently suggest comparable performance, substantial local-level differences remain elusive. To address this gap, we introduce a comparative framework leveraging Bland-Altman agreement analysis, to evaluate per-residue Cα-confidence differences and Per-Residue profiles (PRPs), complemented by Uniform Manifold Approximation and Projection (UMAP). This approach reveals marked localized divergences, particularly within flexible or intrinsically disordered regions, where both predictor choice and single-residue substitutions trigger the largest conformational shifts. We further demonstrate that using reduced homology databases has minimal impact on predicted structural quality, offering computationally efficient alternatives. Collectively, our findings underscore the importance of integrating global and residue-specific evaluations to more accurately assess robustness, agreement, and practical usability across contemporary protein structure prediction methods.

Proteins↗

PubMed2025

Privacy-hardened and hallucination-resistant synthetic data generation with logic-solvers.

MOTIVATION: Machine-generated or synthetic data is a valuable resource for training artificial intelligence algorithms, evaluating rare workflows, and sharing data under stricter data legislations. However, current statistical and deep learning methods struggle with large data volumes, are prone to hallucinating scenarios incompatible with reality, and seldom quantify privacy meaningfully. RESULTS: Here, we introduce Genomator, a logic solving approach (SAT solving), which efficiently produces private and realistic representations of the original data. We demonstrate the method on genomic data, which arguably is the most complex and private information. We benchmark Genomator against state-of-the-art methodologies (Markov generation, Wasserstein Generative Adversarial Network and Conditional Restricted Boltzmann Machines), demonstrating a 40%-530% accuracy improvement and 57%-172% higher privacy. Genomator is also 3-100 times more efficient, making it the only tested method that scales to whole genomes. We show the universal trade-off between privacy and accuracy, and use Genomator's tuning capability to cater to all applications along the spectrum, from provable private representations of sensitive cohorts, to datasets with indistinguishable pharmacogenomic profiles. Demonstrating the production-scale generation of tuneable synthetic genomes hold great potential for balancing underrepresented populations in medical research and advancing global data exchange. AVAILABILITY AND IMPLEMENTATION: Genomator is available at https://github.com/csiro/genomator.

Algorithms↗

PubMed2026

Profiler: an open web platform for multi-omics analysis.

MOTIVATION: High-throughput multi-omics technologies produce increasingly large and heterogeneous datasets that are difficult to analyze without advanced computational expertise. Existing bioinformatics tools are often fragmented or limited to specific omics types, hindering reproducibility and accessibility. There is a critical need for an integrated, user-friendly, and scalable platform capable of supporting multi-omics analyses across different data modalities. RESULTS: We present Profiler, an open-source, modular platform that unifies data import, quality control, preprocessing, statistical testing, machine and deep learning, biomarker discovery, pathway and drug-target enrichment, and survival modeling within a single reproducible environment. Built in Python with Streamlit, Profiler is available as both a web-based platform deployed on high-performance computing and a desktop version for local execution, enabling flexible usage across computational infrastructures. Profiler supports diverse omics modalities, including proteomics, transcriptomics, lipidomics, and electroencephalogram data. Through applications to glioblastoma proteomic, pancancer, and multi-omics datasets, Profiler reproduced known molecular subtypes, revealed potential therapeutic targets, and generated fully traceable analysis reports within minutes. By integrating advanced analytics behind an intuitive interface, Profiler democratizes multi-omics analysis and provides a robust, scalable foundation for systems biology and precision medicine research. AVAILABILITY AND IMPLEMENTATION: Profiler is open-source and freely available via its web platform (https://prism-profiler.univ-lille.fr) and GitHub (web version: https://github.com/yanisZirem/Profiler_v1_requests_datatests, desktop version: https://github.com/yanisZirem/prism-profiler), and archived on Zenodo (DOI: https://doi.org/10.5281/zenodo.17478158).

Software↗

PubMed2026

How negative sampling shapes the performance of transcription factor binding site prediction models.

MOTIVATION: Transcription factors (TFs) are key players in gene regulation and development, where they activate and repress gene expression through DNA binding. Predicting transcription factor binding sites (TFBSs) has long been an active area of research, with many deep learning methods developed to tackle this problem. These models are often trained on TF ChIP-seq data, which is generally seen as only providing positive samples. The choice of datasets and negative sampling techniques is a critical yet often overlooked aspect of this work. RESULTS: In this study, we investigate the impact of different negative sampling techniques on TFBS prediction performance. We create high-quality test datasets based on ChIP-seq and ATAC-seq data, where true negatives can be identified as positions that are accessible but not bound by the TF in question. We then train models using various negative sampling techniques, including genomic sampling, shuffling, dinucleotide shuffling, neighborhood sampling, and cell line specific sampling, simulating cases where matching ATAC-seq data is not available. Our results show that, generally, metrics calculated on training datasets give inflated performance scores. Of the tested techniques, genomic sampling of negatives based on similarity to the positives performed by far the best, although still not reaching the performance of baseline models trained on high-quality datasets. Models trained on dinucleotide shuffled negatives performed poorly, despite being a common practice in the field. Our findings highlight the importance of carefully selecting negative sampling techniques for TFBS prediction, as they can significantly impact model performance and the interpretation of results. AVAILABILITY AND IMPLEMENTATION: The code used in this study is available at https://github.com/NatanTourne/TFBS-negatives (DOI: 10.5281/zenodo.18007567).

Binding Sites↗

PubMed2026

A novel transformer model of protein domains for viral taxonomy classification.

MOTIVATION: Viruses with carefully curated taxonomic assignments (such as those in the ICTV taxonomy) still represent only a small fraction of viruses identified through sequencing data from virome or microbiome projects. It is therefore critical to develop methods that can assign viruses at multiple taxonomic ranks, so that a virus deemed novel at a given rank may still be placed into a higher-level taxon. Sequence-similarity-based approaches can classify viruses that share substantial genomic similarity with known viruses (e.g. those belonging to the same species or genus); however, their performance drops significantly when applied to more divergent viruses. Recent deep learning models, such as ViTax, which utilize DNA language models, aim to address these limitations, but their performance also degrades when applied to novel viruses lacking genus-level similarity to known references. Proteins are more conserved than genomic sequences, and the multiple proteins encoded by a virus can be leveraged to reveal evolutionary relationships among viruses. RESULTS: We propose a new tool, D2T (Domain-to-Taxonomy), that leverages recent advances in protein language models to improve viral taxonomic assignment. D2T represents a virus as a sequence of protein domain tokens and learns a transformer-based model for taxonomic classification. Experiments on multiple closed-set and open-set datasets show that D2T excels at assigning higher-level taxonomic labels (family and above). Furthermore, by combining D2T with Kraken2, which performs well at the genus level, the hybrid method (K+D2T) achieves accurate viral taxonomic classification across multiple taxonomic ranks. AVAILABILITY AND IMPLEMENTATION: D2T is available as a GitHub repository at https://github.com/mgtools/D2T.

Viruses↗

PubMed2026

Artificial Intelligence for Diagnosing Meibomian Gland Dysfunction: A Systematic Review and Meta-Analysis of Diagnostic Test Accuracy Studies.

PURPOSE: To identify, appraise, and synthesize the performance of artificial intelligence-based meibography reading as compared with human graders in diagnosing meibomian gland dysfunction. METHODS: We followed Cochrane methodology and reporting guidelines for diagnostic test accuracy reviews. To assess potential risk of bias and applicability, we used a modified Quality Assessment of Diagnostic Accuracy Studies-2 checklist. We applied bivariate logistic models to estimate summary sensitivity and specificity when appropriate and used the GRADE framework to rate the certainty of the evidence. RESULTS: We identified 14 eligible studies involving 5511 predominantly middle-aged participants (average age: 27-55 years) who were primarily female (≥54.5%). A total of 18,926 meibography images were obtained through noncontact infrared (11 studies) or in vivo confocal microscopy (three studies). Two studies reported external validation of deep learning models, 12 reported internally validated models, and one reported both. All but one study had high risk of bias in at least one domain; 12 studies raised high or intermediate concern about applicability. Based on three external evaluations, the summary sensitivity and specificity for diagnosing meibomian gland dysfunction from normal glands were 97.5% (95% confidence interval: 77.5%-99.8%) and 85.5% (95% confidence interval: 47.3%-97.5%). Sources of heterogeneity in internally validated models included study population, case mix, and others. The overall evidence was very low to low certainty because of imprecision, high risk of bias, and concerns about applicability. CONCLUSIONS: Artificial intelligence-based meibography grading appears less accurate than human graders. Future studies should adopt rigorous designs, including a more diverse participant pool (or image set), and external validation.

Humans↗

PubMed2024

DeepSomatic: Accurate somatic small variant discovery for multiple sequencing technologies.

Somatic variant detection is an integral part of cancer genomics analysis. While most methods have focused on short-read sequencing, long-read technologies now offer potential advantages in terms of repeat mapping and variant phasing. We present DeepSomatic, a deep learning method for detecting somatic SNVs and insertions and deletions (indels) from both short-read and long-read data, with modes for whole-genome and exome sequencing, and able to run on tumor-normal, tumor-only, and with FFPE-prepared samples. To help address the dearth of publicly available training and benchmarking data for somatic variant detection, we generated and make openly available a dataset of five matched tumor-normal cell line pairs sequenced with Illumina, PacBio HiFi, and Oxford Nanopore Technologies, along with benchmark variant sets. Across samples and technologies (short-read and long-read), DeepSomatic consistently outperforms existing callers, particularly for indels.

Journal Article↗

PubMed2025

Carafe enables high quality in silico spectral library generation for data-independent acquisition proteomics.

Data-independent acquisition (DIA)-based mass spectrometry is becoming an increasingly popular mass spectrometry acquisition strategy for carrying out quantitative proteomics experiments. Most of the popular DIA search engines make use of in silico generated spectral libraries. However, the generation of high-quality spectral libraries for DIA data analysis remains a challenge, particularly because most such libraries are generated directly from data-dependent acquisition (DDA) data or are from in silico prediction using models trained on DDA data. In this study, we developed Carafe, a tool that generates high-quality experiment-specific in silico spectral libraries by training deep learning models directly on DIA data. We demonstrate the performance of Carafe on a wide range of DIA datasets, where we observe improved fragment ion intensity prediction and peptide detection relative to existing pretrained DDA models. To make Carafe more accessible to the community, we have integrated Carafe into the widely used Skyline tool.

Journal Article↗

PubMed2025

OpenSpliceAI: An efficient, modular implementation of SpliceAI enabling easy retraining on non-human species.

The SpliceAI deep learning system is currently one of the most accurate methods for identifying splicing signals directly from DNA sequences. However, its utility is limited by its reliance on older software frameworks and human-centric training data. Here we introduce OpenSpliceAI, a trainable, open-source version of SpliceAI implemented in PyTorch to address these challenges. OpenSpliceAI supports both training from scratch and transfer learning, enabling seamless retraining on species-specific datasets and mitigating human-centric biases. Our experiments show that it achieves faster processing speeds and lower memory usage than the original SpliceAI code, allowing large-scale analyses of extensive genomic regions on a single GPU. Additionally, OpenSpliceAI's flexible architecture makes for easier integration with established machine learning ecosystems, simplifying the development of custom splicing models for different species and applications. We demonstrate that OpenSpliceAI's output is highly concordant with SpliceAI. In silico mutagenesis (ISM) analyses confirm that both models rely on similar sequence features, and calibration experiments demonstrate similar score probability estimates.

Journal Article↗

PubMed2025

Integrating Imaging-Derived Clinical Endotypes with Plasma Proteomics and External Polygenic Risk Scores Enhances Coronary Microvascular Disease Risk Prediction.

Coronary microvascular disease (CMVD) is an underdiagnosed but significant contributor to the burden of ischemic heart disease, characterized by angina and myocardial infarction. The development of risk prediction models such as polygenic risk scores (PRS) for CMVD has been limited by a lack of large-scale genome-wide association studies (GWAS). However, there is significant overlap between CMVD and enrollment criteria for coronary artery disease (CAD) GWAS. In this study, we developed CMVD PRS models by selecting variants identified in a CMVD GWAS and applying weights from an external CAD GWAS, using CMVD-associated loci as proxies for the genetic risk. We integrated plasma proteomics, clinical measures from perfusion PET imaging, and PRS to evaluate their contributions to CMVD risk prediction in comprehensive machine and deep learning models. We then developed a novel unsupervised endotyping framework for CMVD from perfusion PET-derived myocardial blood flow data, revealing distinct patient subgroups beyond traditional case-control definitions. This imaging-based stratification substantially improved classification performance alongside plasma proteomics and PRS, achieving AUROCs between 0.65 and 0.73 per class, significantly outperforming binary classifiers and existing clinical models, highlighting the potential of this stratification approach to enable more precise and personalized diagnosis by capturing the underlying heterogeneity of CMVD. This work represents the first application of imaging-based endotyping and the integration of genetic and proteomic data for CMVD risk prediction, establishing a framework for multimodal modeling in complex diseases.

Cardiovascular Disease↗

PubMed2025

Integration of Gene Expression and Digital Histology to Predict Treatment-Specific Responses in Breast Cancer.

Deep learning models applied to digital histology can predict gene expression signatures (GES) and offer a low-cost, rapidly available alternative to molecular testing at the time of diagnosis. We optimized transformer-based models to infer GES results and applied this approach to pre-treatment H&E-stained biopsies from 1,940 breast cancer patients treated with neoadjuvant chemotherapy in clinical trial and real-world cohorts. The most predictive histology-derived GES for pathologic complete response (pCR) in the I-SPY2 trial was validated in four external cohorts: CALGB 40601, CALGB 40603, a trial of durvalumab plus CT, and standard-of-care CT-treated patients from the University of Chicago. Among HER2-negative patients, a transformer-based model trained using a signature composed of estrogen-regulated genes, proliferation, apoptosis, and interferon response genes predicted pCR with an AUC of 0.794, outperforming models based on clinical features alone (AUC 0.704, p = 0.001), pathologist TIL assessment, and a model trained directly to predict response from I-SPY2 cases. Tertiles of this signature stratify patients into clinically relevant groups with increasing likelihood of complete response, with pCR rates ≥50% in the top tertile regardless of treatment or hormone receptor status. Additional transformer-based signature models predicted response to specific therapies (but not chemotherapy alone), including a HER2 signaling signature in IO-treated patients, and a claudin-low signature in bevacizumab treated patients. In HER2- cohorts with available gene expression data and histology, models trained on expression data performed similarly to digital histology predictions, but the combination of gene expression and histology outperformed histology alone. These findings suggest that histology-based GES provides additive information to RNA sequencing data and can inform precision treatment selection across breast cancer subtypes.

Journal Article↗