feature extraction from signals

In clinical proteogenomics, converting raw multi-omics data into reliable, novel biological hypotheses remains a major challenge due to the lack of automated, interpretable frameworks. Method: We propose PROTEUS—the first fully automated hypothesis generation framework that uniformly models the scientific discovery process as an evolvable, interpretable research process graph. It integrates large language models, modular workflow simulation, graph neural network–based representation learning, and an automatic open-scoring mechanism to enable end-to-end analysis of heterogeneous high-throughput data. Contribution/Results: PROTEUS unifies exploratory analysis, statistical testing, and iterative hypothesis generation within a single graph structure, supporting open-science–driven autonomous discovery. Evaluated on 10 public clinical multi-omics datasets, it generated 360 hypotheses; external validation and automated assessment demonstrated significant improvement in the reliability–novelty trade-off. This advances general-purpose AI toward domain-specialized scientific discovery systems.

To address the challenge of fine-grained information extraction from bioinformatics workflows in scientific literature under low-resource conditions, this paper proposes BioToFlow: (1) the first high-quality, manually annotated corpus specifically designed for bioinformatics workflow extraction; (2) a domain-informed named entity recognition (NER) framework that systematically evaluates and integrates four low-resource strategies—SciBERT fine-tuning, few-shot learning, masked language modeling, and knowledge injection; and (3) achieves 70.4 F1 on BioToFlow, matching inter-annotator agreement, with significant performance gains for critical entity types after knowledge enhancement. This work constitutes the first empirical validation of high-accuracy, reusable structured workflow extraction in low-resource settings, establishing a novel paradigm for automated discovery and reuse of computational protocols in scholarly literature.

Medical AI deployment is hindered by insufficient production readiness of machine learning (ML) training pipelines. Method: This paper presents a progressive architectural evolution path—monolithic (chaotic) → modular monolithic → microservices—using SPIRA, a voice-based pre-diagnostic system for respiratory insufficiency, as a case study. It systematically introduces continuous training (CT) and a software-quality-attribute-driven MLOps governance framework tailored to healthcare, integrating modular design, microservice decomposition, and engineered CI/CD pipelines. Contribution/Results: The approach significantly improves pipeline maintainability, fault tolerance, and scalability, enabling stable, iterative evolution of SPIRA. It establishes an “agile ML + robust software engineering” co-design paradigm, delivering a reusable methodology and practical benchmark for engineering medical AI in highly regulated environments.

Bioinformatics education lags behind the growing data-intensive demands of omics research. Method: This study pioneers a systematic integration of the Bioconductor ecosystem—encompassing pedagogical resources, analytical toolchains, and community best practices—into a research-driven, reproducibility-centered teaching paradigm. Leveraging the R/Bioconductor stack, we developed a modular, tiered curriculum spanning beginner to advanced levels, incorporating interactive tutorials (BiocWorkshops), containerized computational environments (Docker/Singularity), and a continuous-integration framework for automated pedagogical assessment. Contribution/Results: The curriculum has been adopted by over 30 universities and training institutions worldwide, yielding significant improvements in learners’ completion rates (+32%) and code reproducibility (+47%) on authentic omics analysis tasks. This work establishes a scalable, open-source, standards-based educational framework for bioinformatics training.

This study addresses the lack of existing multimodal integration methods that translate omics data into testable morphological hypotheses to guide evidence retrieval from histopathology images. The authors propose a closed-loop framework that, for the first time, maps DNA methylation and miRNA features into morphological intent vectors, retrieves relevant pathological regions from structured text via TF-IDF, and employs a cosine similarity gating mechanism to trigger a vision-language model for deterministic refinement. This approach establishes a lexically auditable retrieval-and-verification pipeline, reducing reliance on implicit semantic matching in embedding spaces. Evaluated on the TCGA-BRCA dataset, the method achieves new state-of-the-art performance across multiple clinical tasks—including ER, PR, and HER2 status prediction, molecular subtyping, and risk stratification—significantly outperforming current multimodal fusion and vision-language model baselines.

This study addresses the challenge of effectively integrating heterogeneous proteomic data—specifically, whole-sample mass spectrometry (MS) and multiplexed protein array profiles—in pancreatic cancer research. To overcome the limitations of conventional approaches that naively concatenate multi-source features, the authors propose a novel model fusion framework that explicitly models and leverages the heterogeneity between data sources through a tailored integration strategy. This approach synergistically exploits the complementary strengths of each modality rather than treating them as homogeneous inputs. Experimental results demonstrate that the proposed method significantly outperforms both single-modality models and standard fusion baselines in pancreatic cancer classification, yielding substantial improvements in diagnostic accuracy. The work thus offers a principled and effective paradigm for integrating heterogeneous multi-omics data in biomedical applications.

Biomedical machine learning is often compromised by data leakage arising from repeated measurements, study heterogeneity, batch effects, or temporal dependencies, leading to biased model evaluation. This work proposes a leakage-aware resampling workflow that innovatively integrates leakage-safe data partitioning, training-set-only preprocessing, nested hyperparameter tuning, and post-hoc leakage auditing, culminating in an interactive HTML diagnostic report. Built upon R’s S4 class system, the framework supports classification, regression, and survival analysis tasks while ensuring reproducibility and task-specific evaluation rigor. Simulation studies and multi-study transcriptomic case analyses demonstrate that leakage-preventive pipelines substantially alter model performance and downstream conclusions, underscoring their necessity and practical utility in robust biomedical machine learning.

This work addresses the challenges of model accuracy and reproducibility in medical risk prediction arising from heterogeneous features, scarce samples, and severe class imbalance. The authors propose the yvsoucom-iterkit framework, which innovatively encodes AutoML pipelines as traceable log entities to construct a log-driven, deterministic optimization system. By leveraging configuration-level logs, the method enables component attribution, interaction analysis, and robustness evaluation across random seeds, revealing structured redundancies in the search space and identifying a small set of high-impact components that dominate overall performance. Experiments on the Pima and Stroke datasets demonstrate that optimizing only these critical components achieves Weighted-F1 scores of 0.89 and 0.94, respectively, with ensemble models exhibiting minimal performance variance (0.023–0.026) across random seeds, thereby validating the approach’s effectiveness and stability.

This study addresses the lack of rigorous statistical assessment for the reliability of output structures in complex clustering pipelines that involve multiple data-dependent stages such as anomaly detection, feature selection, and clustering. To bridge this gap, the work systematically applies selective inference to the entire clustering analysis workflow, establishing a statistical framework that enables valid significance testing of final cluster assignments. The proposed method rigorously controls the type I error rate at any pre-specified nominal level and demonstrates strong empirical performance on both synthetic and real-world datasets. By doing so, it provides a principled and reliable foundation for statistical inference in multi-stage, data-driven clustering procedures.

This work addresses the pervasive challenges in bioinformatics tooling—such as fragmentation, complex dependencies, inconsistent documentation, and irreproducible environments—that severely hinder method reuse and adaptation. To overcome these limitations, the authors propose PoSyMed, an open modular platform that integrates biomedical workflows through formalized tool descriptions, containerized execution, a persistent workflow engine, and a conversational interface. Innovatively, a large language model is incorporated as a semantic assistant within a typed, validated, and human-supervised framework to support tool discovery, pipeline construction, and parameter configuration. This design significantly enhances analytical transparency and reproducibility. The platform’s efficacy is demonstrated in representative biomedical use cases, and it has been released as open-source software.