Researchers have launched a new multimodal medical dataset, Observer, capturing anonymized, real-time interactions between patients and clinicians. The dataset links video, audio, transcripts, and electronic health records to study subtleties like body language and environmental factors affecting care.
A new study warns that AI is racing ahead of safety checks in GP clinics, putting patients at risk. The research found that many GPs use AI tools without thorough evaluation or regulatory oversight, carrying risks like automation bias and loss of important social details.
Researchers developed an AI model that can detect coronary microvascular dysfunction using a common electrocardiogram, outperforming previous models in diagnostic tasks. The model can accurately identify a condition often missed in emergency department visits, providing a cost-effective and non-invasive way to diagnose serious heart co...
A team of MIT engineers developed a deep-learning model that predicts how individual cells will fold, divide, and rearrange during a fruit fly's earliest stage of growth. The model achieved 90% accuracy in predicting the movement of 5,000 cells over the first hour of development.
The University of Oklahoma researcher is working on a project using inverse design techniques and AI to create advanced materials that can quickly switch between conductive and insulating states. The goal is to reduce uncertainty in the discovery process and create a scalable material-design methodology.
A Harvard-designed bio-logger captures high-fidelity audio of sperm whale codas, which are later analyzed by machine learning models to uncover structured communication. Recent results show that sperm whales have their own alphabet and use vowels and diphthongs in their language.
A new method has been developed to generate valid confidence intervals for problems involving data that vary across space, helping researchers trust the results of certain experiments. This work can be applied to fields like environmental science and epidemiology.
A new dataset and model improve the efficiency of machine-learning interatomic potentials and their applicability to different chemical elements and material classes. The PET-MAD model uses a compact and denser dataset of 95,595 structures and an original neural network architecture, achieving robust simulations with minimal fine-tuning.
Researchers developed an AI tool to identify patients with undiagnosed Alzheimer's disease using electronic health records, addressing underdiagnosis and healthcare inequities. The model achieved sensitivity rates of 77-81% across diverse populations, promoting fairness while maintaining high accuracy.
A new study used deep learning and large-scale computer simulations to identify structural differences in synthetic cannabinoid molecules that cause them to bind to human brain receptors differently from classical cannabinoids. Researchers found that these substances often trigger the beta arrestin pathway, leading to more severe psych...
Researchers at Florida Atlantic University have developed a deep learning model that detects and evaluates Alzheimer's disease (AD) and frontotemporal dementia (FTD) using EEG brainwave analysis. The model achieved over 90% accuracy in distinguishing individuals with dementia from cognitively normal participants.
Researchers have developed an AI-driven method to uncover genetic interactions driving cancer progression. The approach highlights complex interplay between genes allowing malignant cells to gain momentum and reveals previously hidden cancer drivers.
Researchers at the University of Washington have developed smart headphones that use AI to detect conversation cadence and isolate participants in a noisy soundscape. The prototype was tested with 11 participants, who rated filtered audio more than twice as favorably as baseline sound.
The University of Texas at Dallas has partnered with Tech Mahindra to facilitate collaboration on artificial intelligence (AI) innovation, skill development, and research. The partnership will provide opportunities for students and faculty to advance AI technologies, data science, and cybersecurity.
A breakthrough AI system called OmniPredict can predict human pedestrian behaviors with unprecedented accuracy, revolutionizing self-driving cars and urban mobility. The model combines visual cues with contextual information to anticipate pedestrians' next moves, reducing the risk of accidents and improving traffic safety.
A new advance in EEG imaging technology has improved mapping for epilepsy surgery, providing an accurate and non-invasive method. Pathological HFOs were found to be the most accurate biomarker for identifying epileptogenic brain regions, allowing for precise localization of seizure onset.
The Jackson Laboratory's CARDIOVERSE project uses AI, stem cells, and genetic variation to predict drug safety before human trials. The initiative aims to reduce cardiotoxicity and improve patient stratification in clinical trials.
A Kobe University team has introduced a new method using deep learning for creating tailored simulations that respect physical laws while being computationally efficient. The approach shows superior accuracy in simulating diverse physical systems, including those with chaotic behavior.
Researchers at MD Anderson have made significant discoveries in the treatment of rare bile duct cancers, with zanidatamab showing promising results. Additionally, a study identified RASH3D19 as a target to overcome treatment resistance in KRAS-mutant cancers.
The MIT team developed a new AI-based controller that enables the robot to follow gymnastic flight paths, such as executing continuous body flips. The robot's speed and acceleration increased by 450% and 250%, respectively, compared to previous demonstrations, making it comparable to insects in terms of agility.
Researchers developed instance-adaptive scaling framework that uses process reward model to estimate difficulty of question, enabling LLMs to spend more compute on promising solution paths. This approach achieves comparable accuracy with existing methods while reducing computational cost by up to one-half.
A new study by Cambridge University Press & Assessment and Microsoft Research found that traditional learning activities like making notes remain critical for students' reading comprehension and retention. Note-taking, either alone or combined with large language models (LLMs), was more effective in helping students understand and reme...
The EBRAINS Summit 2025 will bring together experts to assess how neuroscience can drive medical progress, digital innovation, and responsible data use. Preliminary results from the EPINOV clinical trial, integrating virtual brain technology for epilepsy surgery planning, will be presented.
A study compares five DNA foundation language models across 57 diverse datasets to identify their strengths and weaknesses in predicting gene expression, identifying genomic components, and detecting harmful mutations. The findings highlight the importance of selecting appropriate models based on specific genomic tasks.
University of Missouri researchers are combining in-home sensor technology with artificial intelligence to monitor daily changes in ALS patients' health. The system uses machine learning to estimate a patient's score on the ALS Functional Rating Scale Revised, predicting potential problems before they occur.
Researchers at Pusan National University have developed an AI-powered design methodology for gerotor pumps, achieving a 32.3% increase in average flow rate and reducing pressure fluctuation by 53.6%. This innovation has the potential to improve engine durability, lubrication, and cooling, leading to quieter operation and increased reli...
Researchers developed FocalCodec, an audio tokenization method that compresses speech efficiently, preserving its meaning and sound quality. The system was tested with positive results, showing nearly identical reconstructed speech to original recordings.
Researchers developed a machine learning tool, SAMP-Score, to identify compounds that induce senescence in p16-positive cancer cells. The tool successfully identified a promising candidate, QM5928, which triggers senescence without killing cancer cells.
A new AI system can accurately reconstruct hand muscle activity without sensors, enabling precise estimation of fine motor control. The technology has potential applications in sports science, rehabilitation, and human-machine interaction.
A team of researchers has developed a detailed open map of emerging technologies, grouping 23,000 plus technologies into a multi-level map. The Cosmos 1.0 framework uses machine learning to analyze Wikipedia pages, books, and patents.
Researchers developed a reduced order model that accelerates calculations by identifying key features in flow data, enabling faster testing of geometry parameters for efficiency gains. The team plans to make their extensive database and model available online for other research groups.
Large language models can learn to mistakenly link certain sentence patterns to specific topics, causing unexpected failures on new tasks. This shortcoming could reduce the reliability of LLMs used in safety-critical domains like handling customer inquiries and generating financial reports.
Researchers introduced a method to make photonic circuits more adaptable without sacrificing compatibility, enabling the creation of practical photonic quantum neural networks. The approach achieved a classification accuracy above 92 percent in experimental tests, demonstrating its potential.
The National Center for Supercomputing Applications (NCSA) has received the 2025 HPCwire Readers' and Editors' Choice Awards for its outstanding research in artificial intelligence and energy systems. NCSA's premier supercomputing systems Delta and DeltaAI were utilized in two different domains, including a novel AI-based approach to m...
A survey of 134 young European family physicians found moderate overall readiness for AI, with varying levels of knowledge about current applications and usage. The study suggests a need for training and curricula tailored to primary care to address uneven readiness and low day-to-day use.
A new interpretable machine learning framework predicts syngas composition in biomass-plastic co-gasification, identifying key factors controlling product distribution. The model supports improved syngas quality, reduced experimental workload, and more efficient process optimization.
A study at Technical University of Munich found that AI-simplified CT reports reduced reading time to two minutes and improved patient comprehension. Patients rated the simplified texts as more helpful and informative, with 82% finding them easier to read and understand.
Researchers at the University of Utah have developed a new model to predict the snow-to-liquid ratio, which varies widely in the Western United States. By training a random forest model on high-quality data from 14 mountain sites, they were able to explain nearly half of the variability in snow density compared to existing methods.
A new report warns that AI can worsen health disparities in brain disease diagnoses if proper safeguards are not implemented. Researchers call for diverse perspectives, AI education, and strong governance to ensure equitable implementation.
A team of scientists at the University of Tokyo has developed an automated, high-throughput system that uses machine learning to analyze droplets of biofluids for disease diagnosis. The technology relies on imaging drying processes to distinguish between normal and abnormal samples.
PRX Intelligence, a new open access journal from APS, publishes high-impact research on AI and machine learning advancing physical sciences. The journal accepts submissions starting Feb 2026, with flexible article formats and no publication charges for 2026 manuscripts.
A study of 258 published novelists in the UK found that half believe AI will replace their work entirely. The majority also report using AI for non-creative tasks like information searches and editing text written without AI. However, many express concerns about copyright laws not being respected and originality being lost.
Researchers have developed a system called CytoDiffusion that uses generative AI to study the shape and structure of blood cells. The system can accurately identify normal blood cell appearances and spot unusual or rare cells that may indicate disease, outperforming existing systems in tests.
Researchers developed a new approach called SEAL that enables large language models (LLMs) to update themselves permanently. This allows LLMs to learn from user input and improve their performance on tasks like question-answering and pattern-recognition. The technique improves accuracy and enables smaller models to outperform larger ones.
The initiative aims to decipher the basic function of human genomic sequences, enabling personalized diagnosis and therapy. The partnership will analyze genomic regions to identify underlying mechanisms that contribute to disease and uncover potential therapeutic targets.
Researchers at ETH Zurich developed an immersive digital co-pilot to support conservators in restoring historic buildings. The tool uses spatial computing technologies and augmented reality to provide structural analysis insights, enhancing decision-making and disseminating heritage knowledge.
This book explores the impact of decentralized networks on industries like healthcare and supply chains, highlighting the benefits of blockchain technology. It also delves into the synergy between blockchain and emerging technologies like AI and IoT.
Researchers at RIKEN successfully simulated the Milky Way Galaxy with over 100 billion individual stars, far surpassing previous state-of-the-art models. This achievement demonstrates the power of AI-accelerated simulations in tackling complex multi-scale problems in astrophysics and beyond.
This book offers a comprehensive exploration of AI-driven analytics in finance, addressing market prediction, fraud detection, and risk assessment. It also discusses AI applications in healthcare and cybersecurity, including disease classification and biometric identification systems.
A new open-access tool, MOF-ChemUnity, offers a systematic way to organize and synthesize knowledge about metal–organic frameworks (MOFs), enabling the discovery of their potential uses in drug delivery, catalysis, carbon capture, and more. The system creates a unified foundation that both researchers and AI systems can build on, reduc...
The new statistical method adapts to data structure, resisting outliers and providing greater stability on non-Euclidean spaces. This improves the reliability of analysis in areas like medical imaging, computer vision, and machine learning.
The University of Maine has launched internships in AI and digital twins to prepare students for careers in the growing blue economy. Students will work with real-time data, build virtual replicas of ocean structures, and test complex marine scenarios.
Southwest Research Institute uses machine learning to automate calibration of heavy-duty diesel truck emissions control systems, cutting calibration time from weeks to hours. The new method improves system performance while ensuring compliance with upcoming standards.
A recent FAU Engineering study leverages quantum computing to enhance the accuracy of chronic kidney disease (CKD) diagnosis. The research team developed and compared two automated systems: Classical Support Vector Machine (CSVM) and Quantum Support Vector Machine (QSVM). CSVM achieved remarkable 98.75% accuracy, while QSVM reached 87....
The 'Otus' supercomputer provides a solution to pressing challenges through its massive parallel computing capacity, allowing researchers to simulate complex processes, identify patterns, and make predictions about future developments. The system also promotes sustainability with indirect free cooling and renewable energy sources.
The University of Tennessee will lead work in materials and models under a renewed $125M funding for the Quantum Science Center at Oak Ridge National Laboratory. UT's expertise in quantum spin systems will validate quantum-classical computations, while supporting students' involvement in materials science and neutron experiments.
Large language models systematically rate speakers of German dialects less favorably than those using Standard German, associating dialects with negative traits. The bias grows when dialects are explicitly mentioned, and larger models display even stronger biases.
A Dartmouth study finds that AI-powered chatbots can deliver personalized learning to large numbers of students. The researchers created an AI teaching assistant called NeuroBot TA that provides around-the-clock individualized support for students, which they found to be more trusted than general chatbots.
Researchers developed a novel topology-aware multiscale feature fusion network to enhance EEG-based motor imagery decoding. The TA-MFF network achieves excellent classification performance, outperforming state-of-the-art methods by leveraging spectral-topological data analysis-processing and inter-spectral recursive attention.
A new study estimates AI adoption across the US could add approximately 900,000 tonnes of CO₂ annually, a relatively minor increase compared to nationwide emissions. Researchers stress the importance of integrating energy efficiency and sustainability into AI development and deployment to mitigate this environmental impact.