All stories

1313 published stories · page 26 of 88

ScienceEvidence-backed problem

Explainable deep learning improves human mental models of self-driving cars

Self-driving cars increasingly rely on deep neural networks to achieve human-like driving 1-3 . The opacity of these black-box planners makes it challenging to accurately anticipate when they will fail 4-6 , with potentially catastrophic consequences 7-9 . Although research into interpreting these systems has surged, most of it is confined to simulations or toy setups because of the difficulty of real-world deployment 10,11 , leaving the practical utility of these techniques unknown. Here, we introduce the Conce…

Nature

Evidence-backed gain

Integrative bioinformatic analysis delineates a mitochondrial-hematopoietic gene signature for diagnosis and immune characterization in myelodysplastic syndromes

Evidence-backed problem

Physics-Informed Neural Networks Meet Multimodal Large Language Models: Biomechanical Simulation in Aortic Aneurysm

Evidence-backed problem

Machine learning-integrated explainable artificial intelligence for survival prediction in urothelial carcinoma with enfortumab vedotin: an exploratory real-world analysis

Evidence-backed problem

A Real-World Evaluation of Large Language Model-Generated Hospital Courses in Pediatrics

Background Large language model (LLM)-generated hospital courses are increasingly integrated into electronic health records (EHRs), yet their accuracy and safety in pediatric populations remain poorly characterized. Objective To evaluate the accuracy, text quality, and perceived potential harm of EHR-integrated and LLM-generated hospital courses in pediatric inpatient care during early clinical implementation. Methods We conducted a descriptive evaluation from June 10 to August 8, 2025, at an academic freestandi…

Evidence-backed problem

Artificial intelligence and ultra-high performance computing methods and experiments for drug discovery: virtual screening, deep learning, molecular dynamics simulations, ADMET modelling, and experimental validation

Recent years have witnessed considerable progress in computer-aided drug discovery, driven by the incorporation of computational technologies within both academic and pharmaceutical environments. This evolution is marked by a significant accumulation of data pertaining to detailed three-dimensional structural information, ligand properties, and their interactions with therapeutic targets. The augmentation of computational capabilities and the accessibility of extensive chemical libraries containing billions of d…

Evidence-backed gain

Development of a Machine Learning Algorithm for Differential Diagnosis Between Primary Immune Thrombocytopenia and Connective Tissue Disease-Related Thrombocytopenia in Pediatric Patients

Background To develop a machine learning model for early differentiation of primary immune thrombocytopenia (pITP) from connective tissue disease-related thrombocytopenia (CTD-TP) in children presenting with thrombocytopenia. Method A retrospective study was conducted on 387 newly diagnosed children with thrombocytopenia. All patients were clinically diagnosed and divided into a training set and a test set in a 7:3 ratio. Six machine learning algorithms, including XGboost, RF, SVM, LR, GBDT, BPNN, were used to e…

Evidence-backed problem

Machine learning models for early detection of urinary tract infections in kidney transplant patients

Background Renal transplantation is the preferred treatment for end-stage chronic kidney disease but requires lifelong immunosuppression, increasing the risk of infections such as urinary tract infection (UTI). UTI in kidney transplant recipients can lead to serious complications, including acute kidney injury, reduced graft survival, and increased mortality. Machine learning can enhance risk detection accuracy and support proactive management of complications. This study aims to develop a machine learning-based…

Health

SlideChat is a multimodal generative artificial intelligence assistant for whole-slide computational pathology across cancer types

Multimodal artificial intelligence, albeit showing great potential in computational pathology, remains limited to isolated patch-level interpretation and often fails to analyze gigapixel-scale whole-slide images (WSIs) essential for clinical utility. Here we present SlideChat, a multimodal generative artificial intelligence assistant for whole-slide computational pathology across cancer types. SlideChat integrates patch-level and slide-level pathology encoders with a pretrained large language model. Using 274,23…

Health

Ethical Considerations on Artificial Intelligence, Health and Health Care. All That Glisters Is Not Gold

Background Artificial intelligence (AI) is transforming global health care through innovations in deep learning, generative models and agentic AI systems. Traditional reductionist approaches to complex pathophysiological pathways fail to capture the true complexity of disease, motivating the adoption of network medicine (NM), which models biological systems as dynamic, interconnected networks. When combined with AI, NM enables integration of multiomic data and better characterizes disease mechanisms to guide pre…

Health

Clinical Applications of Artificial Intelligence in Cardiac CT: From Coronary CT Angiography to CT-Derived Fractional Flow Reserve and Myocardial Perfusion Imaging

Artificial intelligence (AI) has rapidly transformed cardiac CT, extending its clinical utility from coronary CT angiography (CCTA) to CT myocardial perfusion imaging (CT-MPI). This review outlines the current advances in and future perspectives on AI-aided cardiac CT across anatomical, functional, and prognostic dimensions. In CCTA, AI can automate calcium scoring, vessel segmentation, and plaque characterization, markedly improving workflow efficiency and reproducibility. Deep-learning models can allow accurat…

Climate

Explainable AI-Supported Cyber-Physical Collaboration for Sustainable Manufacturing in Industry 5.0

In this study, we present an innovative approach to the sustainable manufacturing of industrial parts using an explainable artificial intelligence (XAI)- based cyber-physical collaboration system for Industry 5.0. Current cyber-physical human systems (CPHSs) have been found to integrate AI only to a limited extent and often lack explainability. Consequently, there is a need to improve their scalability and flexibility, in keeping with the tenets of Industry 5.0: resilience, long-term viability, and human-centric…

Health

A polyline searching-driven evolutionary AI for disease detection of medical imaging data

Objective. Accurate Ultrasound (US) prostate cancer (PCa) segmentation images hold significant value for organ interventional guidance and clinical disease diagnosis. However, this task still poses substantial challenges. The main obstacles include blurred or incomplete boundaries separating PCa from adjacent soft tissues, shadow artifacts inherent to ultrasound imaging, and drastic inter-patient variations in organ morphological shapes. Approach . To address these issues, our method introduces a novel coarse-to…

Lifestyle

Explainable and domain-adaptive prediction models for refrigerant charging in air conditioning systems within industrial processes

Accurate prediction of refrigerant deficiency in consumer air conditioning (AC) systems is critical for optimizing energy efficiency and operational stability. However, existing data-driven models often suffer from significant performance degradation due to domain shift across different AC types and a lack of explanatory transparency. To address these challenges, we propose AC-RPX (AC-Refrigerant Prediction eXplainable AI), a unified framework that integrates a Domain Encoder augmented with domain-specific token…

Health

Methodological Reporting Quality of Artificial Intelligence-Based Orthopedic Studies: A Literature Analysis

Objective We set out to examine how completely artificial intelligence (AI)-based orthopedic studies report their methods. Methods A PubMed search covering 2023-2025 was run with a predefined strategy. Screening of titles, abstracts and full texts left 280 eligible papers, and 200 of these were drawn at random for scoring. Each paper was rated against a six-item framework built from the TRIPOD-AI, STARD-AI and CONSORT-AI guidelines. Results Mean score across the 200 studies was 4.8±0.9. Twenty-eight percent reac…