Papers for

biomedical software developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Machine learning predicts cancer outcomes across genetic ancestry continuum

Beyond Discrete Ancestry Groups: Distribution-Aware In-Context Learning for Cancer Outcome Prediction Across the Genetic Ancestry Continuum

Abstract: Machine learning models for omics-based prediction of cancer outcomes are commonly built on aggregated data without distinguishing ancestry groups, or under an ancestry-discrete design, which partitions samples into ancestry groups and fits a separate model within each group. Recent evidence shows that this approach is inherently limited: predictive accuracy declines with genetic distance from the training samples, even within a single ancestry group, because the data distribution shifts continuously rather than only between predefined categories. In this paper, we propose distribution-aware in-context learning (DA-ICL) for ancestry-continuum modeling. Our method extends conventional ancestry-discrete modeling to ancestry-continuum modeling, capturing the nonstationary omics-endpoint relationship across the genetic continuum while enabling efficient modeling through an in-context learning foundation model. We provide theoretical analysis demonstrating the superiority of the ancestry-continuum over ancestry-discrete modeling, along with empirical results supporting the proposed framework. Source code is available at https://github.com/ai4pm/DA-ICL-Continuum.

Sat 7 NovArtificial IntelligenceMachine Learning
The gist
Machine learning models often predict cancer outcomes without fully considering genetic ancestry or by grouping people into broad ancestry categories. The authors found that prediction accuracy drops as a person's genetics differ more from those in the training data, even within the same ancestry group. They propose a new method called distribution-aware in-context learning that models genetic ancestry as a continuum rather than fixed groups. This approach better captures how genetic differences influence cancer outcome predictions and improves accuracy. They provide theoretical reasoning and experimental evidence supporting their method.
Open → 2611.10464v1

Meta-initialization speeds up deformable medical image registration

Meta-Initialization of Implicit Neural Representations to Accelerate Deformable Medical Image Registration

Abstract: Implicit neural representation (INR) methods for deformable medical image registration require the network parameters to be optimized for each new image pair, making adaptation speed a practical limitation. This work investigates whether early adaptation can be accelerated using an initialization learned from a collection of previously solved registration tasks. We examine two schemes: Reptile meta-initialization and STRAINER transfer of the shared component of a coordinate network. KAN-IDIR is used as the base INR displacement-field model. Experiments are conducted on OASIS-1, with 364 consecutive pairs used for training and 49 pairs for test-time adaptation. Registration quality is evaluated using mean Dice and HD95, together with the fraction of voxels with a negative Jacobian determinant at iteration 1500. In the configurations studied, both schemes yield higher Dice and lower HD95 at early iterations than random initialization. Reptile numerically outperforms STRAINER under a small iteration budget, although their pretraining budgets are not matched. As the number of iterations increases, the differences between the schemes shrink. STRAINER pretraining takes 39-40 minutes, compared with 44.98 hours for Reptile.

Tue 27 OctComputer Vision and Pattern RecognitionMachine Learning
The gist
Registering medical images, like matching one brain scan to another, often requires slow tuning for each new image pair. The authors study ways to start this tuning process faster by learning from many past registrations. They try two techniques that prepare the neural network before seeing new images, helping it adapt quicker. Their tests show both methods improve early results, with one being faster to train than the other.
Open → 2610.37965v1

Ai agents struggle to exactly reconstruct sequencing workflows and molecular structures

Evaluating and improving AI agents for sequencing protocol understanding

Abstract: Existing scientific-agent benchmarks rarely evaluate whether agents can reconstruct the exact molecular structures and generation workflows specified by sequencing protocols, a capability critical for analyzing customized sequencing assays. We introduce LibStructBench, a document-grounded benchmark for reconstructing sequencing-library structures and molecular workflows. Our experiments show that frontier agents often produce approximately correct reconstructions but remain unreliable at exact reconstruction, particularly in molecular-state representation and workflow connectivity. Switching the underlying language model produced larger average performance differences than switching agent harnesses, with a statistically significant difference only for molecular-state reconstruction. Additional knowledge improved oligo-family recovery in a small pilot without improving mean molecular-workflow scores. Autonomous accumulation of procedural rules and cross-protocol examples showed non-monotonic validation performance and no mean held-out molecular-workflow improvement in the tested trajectory. These findings highlight the need for more controlled and verifiable approaches to scientific reasoning about molecular processes in complex experimental protocols.

Tue 27 OctArtificial Intelligence
The gist
Sequencing protocols describe how to prepare molecules for genetic analysis, and it's important to understand exactly how molecules are arranged and processed. The authors created a new test called LibStructBench to see if AI agents can accurately rebuild these molecular structures and workflows from documents. They found that while current AI can roughly guess the setups, they often make mistakes in important details like molecular states and connections. The study shows that just changing the underlying AI language model affects performance more than changing the way the AI is organized. Adding extra knowledge helped in some small ways but didn’t improve overall workflow understanding.
Open → 2610.37851v1

Random jump forests estimate disease progression without markov limits

Random Jump Forests

Abstract: Random jump forests are introduced as a nonparametric tool for estimating conditional cumulative transition rates and occupation probabilities in multi-state models. The main idea is that a single tree partitions the covariate space for all transitions simultaneously, using partially observed trajectories, and with Nelson--Aalen estimators at the terminal nodes. Occupation probabilities are recovered through the product integral without requiring the Markov assumption. Uniform consistency and a functional central limit theorem for the estimators of cumulative rates can be established under the assumption that the covariate vector takes finitely many values, and under suitable splitting conditions. In particular, pointwise confidence intervals and simultaneous bands are available in the Markov case. Simulation studies show a predictive accuracy close to the correctly specified benchmark models and marked improvements when benchmarks are misspecified, with further illustration on four biomedical data sets. In the coverage study, confidence bands attain near-nominal coverage in several settings, with pooling of heterogeneous rates being the main source of possible undercoverage. The method, including prediction assessment and variable importance, is implemented in the R package JumpForests.

Fri 23 OctMachine Learning
The gist
Predicting how diseases or conditions change over time can be tricky, especially when the usual assumptions about patient transitions don't hold. The authors present random jump forests, a method that looks at patient data all at once and estimates how likely different disease stages are over time without needing strict assumptions. Their approach can give reliable predictions and confidence intervals to understand uncertainty. It works well in simulations and real medical datasets, and they've made it available as a tool in the R programming language.
Open → 2610.32498v1

Ki67 cell typing from images using pretrained embeddings and clustering

K-CLUE: A Segmentation-to-Clustering Framework for Ki67 Cell Typing

Abstract: Ki67 is a key marker of tumor proliferation in breast cancer, but its quantification from immunohistochemistry images normally depends on supervised classifiers trained per dataset. We examine the role of frozen pretrained visual representations in annotation-free Ki67 cell typing. To this end, we use K-CLUE, a framework that segments individual nuclei, represents each nucleus with a pretrained encoder, reduces the resulting embeddings with principal component analysis, and separates Ki67-positive from Ki67-negative nuclei using k-means clustering, without task-specific training or manual annotations. Clustering the raw components is highly sensitive to the choice of encoder: across seven encoders, the Ki67-positive F1-score on matched MIST nuclei varies by 44 points, although Ki67 status remains linearly decodable from every representation, with supervised probe AUCs above 0.98. Analysis of the embedding spaces shows that clustering succeeds when DAB-related variation occupies the leading directions of variance. Whitening the retained principal components removes this dependence, reducing the spread across encoders to 1.00 F1-score point. Handcrafted color representations are also competitive, confirming that Ki67 status is strongly reflected in chromatic information, but they require an explicit choice of color space. In end-to-end evaluation, K-CLUE reaches F1-scores of 85.00% and 81.91% for Ki67-positive and Ki67-negative nuclei on MIST, and 79.59% and 81.16% on BCData, outperforming DeepLIIF and IHCScoreGAN on both datasets. These results show that pretrained representations can support annotation-free Ki67 typing when their variance structure is accounted for.

Fri 23 OctComputer Vision and Pattern Recognition
The gist
Quantifying Ki67, a protein marker for tumor growth in breast cancer, usually requires training on labeled images for each dataset. The authors present K-CLUE, a method that identifies and classifies cells stained for Ki67 without needing any manual labeling or retraining. It first segments nuclei in images, extracts features using a pretrained model, then groups the cells using clustering to distinguish Ki67-positive from negative. Their study shows that by adjusting how features are processed, this clustering can be made reliable across different pretrained models, and their method performs well on real breast cancer image datasets.
Open → 2610.32493v1

AutoMIL automates cancer image analysis with high accuracy

AutoMIL: Automated Multiple Instance Learning for Whole-Slide Image Classification

Abstract: Developing multiple instance learning pipelines for the classification of whole-slide images involves many interdependent design decisions and expertise in both pathology and machine learning. We present AutoMIL, an open-source framework that applies principles from the field of automated machine learning to tackle this task by automating integral pipeline steps, such as preprocessing, training and evaluation. In a benchmark evaluation with default settings, AutoMIL reached ROC-AUC scores of 0.94 on UBC-OCEAN, 0.99 on PCam and 0.93-0.96 on a TCGA-NSCLC subset, comparable to specialised pipelines in related work. Ensembling mainly improved performance on imbalanced data, showing that an automated pipeline must adapt to dataset characteristics. AutoMIL thus provides an accessible baseline for whole-slide image classification without manual pipeline configuration.

Fri 23 OctComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Classifying whole-slide medical images is very complex and often needs experts in both medicine and machine learning. The authors developed AutoMIL, a tool that automates this process, handling steps like image preparation and training without manual setup. AutoMIL performed nearly as well as specialized expert systems across several datasets, showing it can be a reliable baseline for analyzing pathology images. The tool also can adjust to different data characteristics, which improves results on unbalanced datasets.
Open → 2610.32452v1

MultiPASE aligns and analyzes 3D genome shapes across conditions

MultiPASE: Shape-Aware Comparative 3D Genome Analysis

Abstract: 3D genome structures reconstructed from Hi-C data provide valuable insights into chromatin organization, but comparing them across conditions and time points remains challenging because of noise, variability, and arbitrary orientation. State-of-the-art comparison approaches often rely on contact-matrix-based summaries, which do not preserve the spatial geometry of the reconstructed structures and make integration with gene expression and chromatin accessibility difficult. There is therefore a need for robust structural alignment and spatial abstraction methods that support integrated analysis across conditions and time points. We present MultiPASE, a computational pipeline for comparative analysis of reconstructed 3D genome structures represented as gene-level spatial point clouds. MultiPASE aligns point clouds into a shared coordinate frame using coarse-to-fine rigid registration with PCA-based axis disambiguation and robust ICP refinement, then summarizes spatial organization using two complementary shape abstraction methods that produce stable, nested, multi-scale representations from dense core regions to broader structural extent. Using Hi-C-derived reconstructions from viral infection experiments, we show how the approach reveals condition-dependent structural changes that are difficult to interpret from raw 3D structures alone. By integrating spatial structure with gene expression and chromatin accessibility, MultiPASE supports interactive analysis of spatial organization and gene relocation patterns. A quantitative validation, reported in the Supplementary Materials, shows that the proposed shape abstractions preserve non-convex features, remain stable across samples, and that the alignment pipeline is necessary for meaningful comparison. A usage case and a usability evaluation with eight participants demonstrate the pipeline's effectiveness for comparative structural analysis.

Fri 16 OctHuman-Computer Interaction
The gist
3D genome structures help scientists understand how DNA is arranged in cells, but comparing these structures over time or between conditions is tricky because they can be noisy and oriented differently. The authors developed MultiPASE, a tool that carefully lines up these 3D structures and breaks them down into simpler shapes to compare them better. This method helps reveal changes in genome organization due to viral infection that are hard to see otherwise. It also lets users link the 3D structure with gene activity and DNA accessibility for a fuller picture.
Open → 2610.22786v1

Trajectory guided steering reduces bias in clinical ai predictions

Trajectory-Guided Steering: Mitigating Spurious Feature Bias in Clinical Classification

Abstract: Large language models rely on spurious features in clinical classification: swapping a single gender token in clinical notes alters gender-irrelevant diagnoses in up to 53.5% of cases. Existing inference-time corrections either need counterfactual data built around the specific spurious feature or behave unpredictably. We observe that how a model's preference between answers evolves across layers, its prediction trajectory, indicates for each document whether a correction is likely to help. We propose Trajectory-Guided Steering (TGS), which decides from the trajectory alone whether to correct each prediction and in which direction, with no counterfactual data and no knowledge of the spurious feature. On 16 clinical classification tasks from i2b2, TGS fixes 17.7% of errors while breaking only 1.3% of correct predictions and stays net-positive across all tested hyperparameter settings; all baselines are net-negative or near zero. The same corrections reduce reliance on the spurious feature: on the most affected diseases, gender-swap inconsistency falls sharply (hypertension 53.5% -> 38.0%, diabetes 35.1% -> 21.5%). Applying the correction to every document instead raises breaks on correct positive predictions from 0.1% to 6.4%, so per-document selection is essential. TGS also transfers to three model families and three other clinical and biomedical datasets.

Thu 15 OctComputation and Language
The gist
Clinical AI models sometimes make mistakes because they rely on unrelated details like a patient's gender when diagnosing. The authors found that by watching how a model’s guesses change inside it, they can tell if correcting these mistakes will help. Their method, called Trajectory-Guided Steering, decides whether and how to fix each prediction without needing special data about the bias. This method improves accuracy while reducing bias related to gender in medical diagnoses.
Open → 2610.22618v1

Synergistic multimodal learning improves clinical prediction from medical data

SYNPRED: A Synergistic Approach to Multimodal Learning for Clinical Prediction

Abstract: Multimodal medical data captures complementary aspects of disease. It provides a comprehensive basis for diagnosis and the identification of biological associations linked to progression or treatment response. Established contrastive learning approaches exploit shared information across modalities, but fall short in exploiting complementary, modality-specific signals relevant for a task. Here, we propose a synergistic multimodal variational autoencoder (VAE) framework that separates task-relevant unimodal and multimodal information from instance-specific variation task-irrelevant, allowing complementary signals across modalities to be leveraged for prediction. The method yields benefits even if modalities are missing at test time. We evaluate the method on a treatment response prediction task using medical imaging and RNA data. Empirical comparison with alignment-based and multimodal VAE state-of-the-art approaches demonstrates improved multimodal prediction performance, and consistent benefits for prediction from a single modality after multimodal training. The full code can be found here: https://github.com/cirmuw/SYNPRED.

Tue 13 OctMachine LearningComputer Vision and Pattern Recognition
The gist
Combining different types of medical data can help doctors predict how diseases respond to treatment. Existing methods focus on shared information between data types but miss unique clues each type offers. The authors created a new approach that separates the important signals from noise in each data type and uses them together to make better predictions. This method works well even if some data types are missing and improves accuracy for both combined and individual data sources.
Open → 2610.18110v1

Biomedical entity linking improves without training on special data

Training-Free BioNEL-RAG: Advancing Biomedical Entity Linking through Retrieval-Augmented Generation and Large Language Model

Abstract: Biomedical Named Entity Linking (BioNEL) is the task of mapping mentions extracted from biomedical texts to canonical entities or concepts in the standardized knowledge base. Although existing methods have achieved remarkable performance in mapping accuracy, most rely on training with domain-specific datasets, which fundamentally limits their cross-domain adaptability. To address this limitation, we propose BioNEL-RAG, a novel training-free framework that combines retrieval-augmented generation (RAG) with large language models (LLMs). Within this framework, the RAG module focuses on retrieving candidate entities, while the large language model module plays the role of selecting the most appropriate candidate for entity normalization, which is based on the analysis of the mention's context, with the aid of its remarkable text comprehension capabilities. Experimental results demonstrate that BioNEL-RAG achieves performance comparable to existing methods without requiring training on subdomain datasets, whereas these methods necessitate fine-tuning on specific subdomains such as genes, diseases, or chemical compounds.

Sat 10 OctComputation and Language
The gist
Linking names of things like genes or diseases in medical texts to their official entries is important but usually needs lots of specific training. The authors propose a method called BioNEL-RAG that doesn't require extra training for different medical topics. It works by first finding possible matches and then using a large language model to pick the best one based on the text around the mention. This approach performs just as well as other methods that need tailored training for each topic.
Open → 2610.14789v1

Kernel ensemble improves patient survival predictions in healthcare data

KESurv: A Kernel Ensemble Method for Patient-Specific Survival Prediction

Abstract: Predicting patient-specific survival functions is crucial for clinicians in making informed decisions about patient care and treatment strategies. Among the various models available, the Survival Forest has demonstrated significant effectiveness in numerous scenarios. In this work, we propose an ensemble method that leverages the strengths of the Survival Forest as the master model, complemented by several base models. This ensemble incorporates the Beran estimator, a type of kernel estimator, to enhance predictions of patient-specific survival curves. We evaluated the performance of our proposed model using four distinct healthcare datasets. The results highlight the superiority of our ensemble method over baseline models in both calibration and ranking across most datasets. The findings suggest that our approach offers a more accurate and reliable estimation of patient-specific survival functions, providing a valuable tool for clinical decision-making.

Mon 5 OctMachine Learning
The gist
Predicting how long a patient might survive is important for doctors to decide the best treatments. This paper introduces a new method that combines different prediction models, including one called Survival Forest, to make better predictions for individual patients. The authors tested their method on four healthcare datasets and found it generally performs better than other approaches. This could help doctors have more accurate survival estimates tailored to each patient.
Open → 2610.06434v1

Agentic models discover patient drug response equations faster

APOD: reasoning-guided agentic population ordinary differential equation discovery for pharmacological digital twins

Abstract: Establishing ordinary differential equations (ODEs) describing population data is a fundamental part of mathematical modeling in pharmacology, crucial to developing digital twins. However, doing so from sparse, noisy data is a slow, expert-driven task. Existing automated methods either search a restricted model space or ignore population inter-individual variability. Here we introduce APOD (Agentic Population ODE Discovery), a language-model agent that iteratively reasons over biological knowledge and fit diagnostics in an open-ended search space to discover a population digital twin (PDT), i.e., a shared ODE system with between-subject variability. On synthetic pharmacokinetic and tumor-dynamics benchmarks, APOD recovered ground-truth structures in 94-100\% of runs, 12-fold faster in median than an established library-based search. On real cohorts it converged to valid structures, and proposed a PDT of radioligand-therapy-induced platelet dynamics that predicts thrombocytopenia from first-cycle data and simulates alternative dosing schedules that lower the predicted risk of toxicity.

Mon 5 OctArtificial IntelligenceMachine Learning
The gist
Finding mathematical equations that describe how groups of patients respond to drugs can take a lot of time and expert work, especially when data is limited or noisy. The authors created APOD, a computer program that uses reasoning and biological knowledge to discover these equations automatically and more quickly. They tested APOD on simulated and real medical data, and it accurately found the right equations and predicted drug effects for patient groups. This approach could help doctors better understand and tailor treatments using computer models called digital twins.
Open → 2610.06227v1

Biomedical entity recognition improves with code instruction tuning and voting

Enhancing Biomedical Named Entity Recognition via Multiple Programming Languages Instruction Tuning and Ensemble Method

Abstract: Instruction tuning has become a common paradigm for applying large language models (LLMs) to biomedical named entity recognition (BioNER). However, existing instruction-tuning approaches still face two key challenges. First, conventional natural-language instructions typically serialize BioNER annotations as flat textual outputs, providing limited structural constraints for typed entity extraction. Second, high-quality biomedical annotations are limited, and learning from a single serialized output form may restrict structural diversity and reduce model robustness. Although external biomedical knowledge can be introduced to alleviate data scarcity, it often requires costly resource construction. To address these challenges, we propose MITE, a Multiple Programming Languages Instruction Tuning and Ensemble method for BioNER. MITE reformulates BioNER as a structure-to-structure generation task by representing both instructions and entity outputs in code-formatted representations. Specifically, each training instance is transformed into multiple programming-language formats, including Python, C++, and Java, while preserving the same underlying entity semantics. These language-specific representations provide structurally diverse supervision without requiring external biomedical knowledge or additional annotations. During inference, MITE aggregates predictions from different code formats through an entity-level voting strategy, reducing language-specific prediction variance and improving robustness. Experiments on six widely used BioNER datasets demonstrate that MITE consistently outperforms representative BERT-based and LLM-based baselines and exhibits strong cross-dataset generalization. Ablation and parameter analyses further verify the effectiveness and robustness of the proposed components.

Fri 2 OctComputation and Language
The gist
Biomedical named entity recognition means finding important words like diseases or drugs in medical text. The authors found that teaching AI models using examples written in computer code rather than plain text helps the AI understand the structure better. They used multiple programming languages like Python and Java to give different views on the same data, making the AI more reliable. When the model makes predictions, it combines results from all code styles to get better answers. This approach worked better than other methods on many tests with medical texts.
Open → 2610.02949v1

RareDx improves rare disease diagnosis using controlled evidence and knowledge graphs

RareDx: Controlled Knowledge Integration and Graph-Grounded Policy Optimization for Rare-Disease Diagnosis

Abstract: Rare-disease diagnosis is a long-tail reasoning problem: phenotypes are incomplete, individual disorders are sparsely documented, and relevant evidence is distributed across ontologies, gene annotations, and biomedical text. Language models consequently favor common conditions, miss rare candidates, or produce plausible but invalid names. We introduce RareDx, which couples controlled evidence use with knowledge-graph-grounded policy optimization. RareDx-Harness normalizes heterogeneous records into one ranked-diagnosis task and compares direct inference, static retrieval, adaptive tools, and structured phenotype-gene-disease reasoning over a shared knowledge layer. The training pipeline combines Top-10 post-training with RareDx-KGPO, our knowledge-graph-grounded policy optimization method. Its reward projects predictions into a canonical disease graph and integrates curated graded relevance, ontology proximity, biomedical similarity, and phenotype consistency. Vocabulary and output-budget constraints prevent dense partial credit from rewarding fabricated or overlong differentials. Across eight benchmarks, the complete RareDx system centered on Qwen3.5-9B reaches 38.34 macro Hit@10, 1.60 points above GPT-5.5 under the archived protocol; a disjoint validation-selection audit retains a 6.80-point routing gain over Direct on held-out cases. The 27B system reaches 23.53/36.56/40.76 at Hit@1/5/10. Controlled ablations show that retrieval is not uniformly helpful and that controlled routing is central to the gain. These results indicate that structured medical knowledge can turn a compact model into a competitive diagnostic ranker across heterogeneous long-tail settings in clinical practice.

Mon 28 SeptArtificial Intelligence
The gist
Diagnosing rare diseases is hard because relevant information is scattered and rare conditions have little data. The authors developed RareDx, a system that combines careful use of evidence and a knowledge graph to better rank possible diagnoses. This approach helps the system avoid common mistakes like favoring common diseases or inventing incorrect ones. Tests show RareDx improves diagnostic accuracy over other models, especially for rare diseases.
Open → 2609.35549v1

Whole-slide image analysis models tumor microenvironment interactions dynamically

Modeling Whole-Slide Images as Dynamic Tumor Microenvironment Fields

Abstract: Due to the gigapixel-scale nature of whole-slide images (WSIs), weakly supervised WSI analysis is commonly formulated as a multiple instance learning (MIL) problem, where patch-level features are aggregated into slide-level representations. However, diagnostic and prognostic evidence often arises from spatially coherent tumor microenvironment regions and their interactions, rather than isolated patches alone. Existing patch-level or static region-based methods usually overlook how tissue regions should be adaptively formed and subsequently evolved through microenvironment interactions across heterogeneous boundaries. In this paper, we propose Concept-Guided Tumor Microenvironment Evolution (TMEvolve), a reaction-diffusion-inspired framework that models WSIs as latent tumor microenvironment fields over discrete patch graphs. TMEvolve instantiates this view as a learnable graph-discretized evolution process over patch neighborhoods. It first forms adaptive soft tissue regions as coherent microenvironment units, then performs pseudo-time evolution through two complementary local dynamics: intra-region diffusion, which stabilizes latent states within coherent tissue compartments, and concept-guided boundary flux, which propagates visual feature signals and language-derived concept signals across heterogeneous region interfaces. The evolved microenvironment regions are finally aggregated for slide-level prediction. We evaluate TMEvolve on six datasets across three weakly supervised WSI tasks: survival prediction, gene expression prediction, and histological subtype classification. TMEvolve consistently improves over representative MIL methods, pathology foundation models, and concept-guided baselines. Ablation studies and visualizations further support the effectiveness and interpretability of TMEvolve, highlighting the value of dynamic region modeling and boundary interaction.

Mon 28 SeptComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Whole-slide images of tissue samples are huge and hard to analyze all at once. The authors designed a new method, TMEvolve, that groups nearby tissue patches into regions and models how these regions interact and change over time, like a dynamic environment. This helps capture important spatial relationships that simpler methods miss. Their method improves predictions about cancer survival, gene activity, and tissue types compared to older techniques.
Open → 2609.34451v1

Histology predicts gene expression using spatial patterns and low-rank programs

MoSPR: Histology-to-Gene Expression Prediction with Morpho-Spatial Macrostates and Low-Rank Molecular Programs

Abstract: Predicting molecular profiles from histopathology remains challenging because whole-slide images contain spatially organized, heterogeneous tissue patterns, while gene expression comprises thousands of correlated targets. We introduce MoSPR (Morpho-Spatial Program Regression), a linear framework that couples an adjacency-informed histology representation with a low-rank molecular basis. MoSPR clusters frozen patch embeddings into morphology microstates, aggregates their spatial adjacencies across the training cohort, and groups microstates with similar adjacency patterns into shared macrostates. Each slide is then represented by global morphology and macrostate-specific deviations, which are linearly mapped to coefficients of a training-derived low-rank gene-expression basis. Across three cancer cohorts from The Cancer Genome Atlas, MoSPR achieves the highest mean gene-expression prediction scores among all evaluated methods. Without pathway-level supervision, pathway scores derived from its predicted expression profiles rank first in eight of nine comparisons across three pathway collections. Ablation studies on the breast cancer cohort show complementary gains from adjacency-derived macrostate representation and low-rank molecular prediction. Moreover, with half of the training data on this cohort, MoSPR exceeds the full-data gene-prediction score of the strongest competing baseline. Finally, its linear formulation enables exact decomposition of each predicted expression profile into global and macrostate-specific molecular contributions, providing an interpretable link between spatially coherent macrostate regions and their associated molecular programs. Our code is available at https://github.com/Radisen-Panthera/MoSPR.

Mon 28 SeptArtificial IntelligenceComputer Vision and Pattern RecognitionMachine Learning
The gist
Predicting gene activity from tissue images is hard because tissues have many small, complex patterns and genes are linked in many ways. The authors created MoSPR, a method that groups tiny tissue patterns by their surroundings to form bigger, meaningful regions. These regions then help predict gene activity accurately using a simpler set of gene patterns. MoSPR works better than other methods on several cancer datasets and can clearly explain how specific tissue areas relate to gene activity.
Open → 2609.34280v1

Cytospm improves detection of diverse cell types in cytopathology images

CytoSPM: Open-Vocabulary Cytopathology Detection with Structured Prompt Bank

Abstract: Cytopathology detection requires open-vocabulary recognition because cellular categories are fine-grained, long-tailed, and continuously evolving across different organ systems. However, existing cytology detectors are mostly single-domain and closed-set, and there is still no unified benchmark for evaluating open-vocabulary cytopathology detection. We present PentaCyto, a multi-domain benchmark covering cervical, urinary, respiratory, serous fluid, and thyroid cytology, with 24 base categories and 9 held-out novel categories. Each category is associated with structured cytomorphology prompts that describe diagnostic morphological attributes and provide clinically grounded textual knowledge. We further propose CytoSPM, an efficient detector based on a decoupled two-stage design. It first extracts reusable class-agnostic visual representations, and then performs class-aware structural prompt matching with class names and cytomorphology prompts. On PentaCyto, CytoSPM outperforms existing methods in novel-category detection and open-vocabulary detection while maintaining efficient inference.

Fri 25 SeptComputer Vision and Pattern Recognition
The gist
Detecting different types of cells in medical images of body fluids is hard because there are many tiny and similar-looking cell types which keep changing depending on the organ. The authors created a new test called PentaCyto that covers five different body fluids and groups cell types into known and new ones. They designed a tool called CytoSPM that first finds general visual features in images and then matches them with detailed text descriptions of cell shapes and features. CytoSPM works better than previous methods at spotting new cell types accurately and quickly on this test.
Open → 2609.31314v1

Biomedical data clusters improved by causal and flexible dependence modeling

Copula Adapted Directed Acyclic Graph for Cluster Representation of Biomedical Data

Abstract: Diagnostic errors and mislabeling are common in biomedicine, which compromise the reliability of predictive models and data-driven outcomes. Stratifying unlabeled biomedical data based on complex relationships between features eliminates the need for data labels and overcomes the limitations of supervised learning. Traditional clustering methods assume restrictive data distributions, making them suboptimal for capturing complex dependencies in high-dimensional biomedical data. This paper introduces a novel cluster-friendly data presentation framework that integrates the non-Gaussian and non-linear feature dependence of copula models with an ensemble of causal structure discovery (CSD) methods based on Directed Acyclic Graphs (DAGs). While copulas model flexible multivariate distributions by relaxing assumptions related to multivariate normality, linear dependence, and symmetric relationships, an ensemble of DAG-based CSD methods identifies stable causal relationships between features. When clustered using K-means, the new data representation obtained by the proposed copula-adapted DAG (CopDAG) ranks first among the 12 methods in normalized clustering accuracy and adjusted Rand index across 16 biomedical datasets. Our CopDAG method predicts ground-truth class labels directly from feature relationships without data annotations and supervised learning, while also providing cluster visualizations and explainable causal structures of the biomedical data features.

Mon 14 SeptMachine Learning
The gist
Biomedical data can be complicated and often lacks clear labels, making it hard for computers to group similar patients or conditions accurately. The authors developed a new way to look at the data by combining methods that model complex relationships and find cause-effect links between features. This approach helps group data more accurately without needing labels, providing clear visualizations and insights about what features matter. Their method outperformed many other techniques on multiple biomedical datasets.
Open → 2609.16240v1

Conversational agent improves querying clinical trial information

ClinAgent: A ReAct-Based Agent for Conversational Access to Clinical Trial Information

Abstract: Querying clinical trial registries remains a manual and error-prone process, requiring researchers to navigate large volumes of semi-structured data without support for natural language interaction or cross-source synthesis. To address this, we introduce ClinAgent, a conversational system based on agentic Retrieval-Augmented Generation (RAG) that enables clinicians and researchers to query clinical trial information in plain language and receive grounded, up-to-date responses across multi-turn interactions. The system centers on a Large Language Model (LLM) agent following the ReAct paradigm, which iteratively reasons over queries, selects among a set of integrated tools, and refines its actions based on intermediate outputs. These tools include a ClinicalTrials.gov search interface, a PubMed module, and a Python-based analyzer operating on a locally cached structured dataset of clinical trials. We evaluate the system using a three-phase framework assessing operational effectiveness, planning quality, tool-use efficiency, and expert qualitative judgments, comparing three LLM backends: Gemini 3.0 Flash and two variants of DeepSeek V3.2 (thinking and non-thinking). Results reveal complementary strengths, with DeepSeek (thinking mode) excelling in planning quality, while Gemini achieves the highest overall performance and strongest expert ratings. Overall, our findings highlight the potential of agentic AI systems to improve the accessibility and synthesis of clinical trial information, supporting more efficient and user-centered biomedical research workflows.

Sat 12 SeptArtificial IntelligenceComputation and LanguageInformation Retrieval
The gist
Finding detailed information about clinical trials is often hard because the data is scattered and technical. The authors created ClinAgent, a system that lets users ask questions in plain language and get clear, up-to-date answers about clinical trials. It uses an advanced AI approach that combines reasoning with tools to search databases like ClinicalTrials.gov and PubMed. This makes it easier for doctors and researchers to find and understand clinical trial data through a conversation-like interface.
Open → 2609.13860v1