Researchers at UCSF and Wayne State University found that generative AI tools can perform orders of magnitude faster than human teams in analyzing health data. Junior researchers paired with AI generated viable prediction models in minutes, outperforming experienced programmers in hours or days.
SeaCast consistently outperforms the Copernicus operational model over a 10-day forecast horizon and extends predictions to 15 days, generating forecasts in just 20 seconds using a single GPU. This advancement enables rapid 'what-if' scenario testing and probabilistic ensemble forecasts.
This study developed and evaluated an automatic method for lung nodule detection and classification using a CNN-based architecture on the LIDC-IDRI database. The proposed method achieved high sensitivity and accuracy, with competitive performance compared to recent studies.
A University of Houston professor has found that tree-like thin films release heat at least three times better than traditional methods, enabling more efficient cooling in AI data centers. The discovery demonstrates the power of physics-aware AI design for validating high-impact cooling solutions.
Researchers at Nagoya University developed an AI system called YORU that recognizes animal behaviors with over 90% accuracy. The system combines real-time video capture with optogenetics to selectively target brain cells driving specific behaviors, offering a major breakthrough in social behavior studies.
Researchers at UC San Diego developed a new training method for AI systems to improve their performance in solving complex problems that require both text and image interpretation. The approach evaluates the quality of training data and grades models based on their logical reasoning, reducing the risk of incorrect interpretations.
Researchers at Oregon State University have developed a deep learning-based model for rapid bacterial contamination detection, eliminating misclassifications of food debris. The enhanced model can reliably detect bacteria in three hours and has the potential to prevent outbreaks and protect consumer health.
Researchers have developed a passive, solar-powered orbital data center that can scale AI computing and reduce environmental impact. The system leverages decades of research on 'tethers' and could host thousands of computing nodes to replicate terrestrial data centers.
A recent study from Binghamton University School of Management reveals that focusing on human-robot collaboration can generate additional economic value and improve a company's ability to capture a greater share of the competitive market. By leveraging robots in collaborative settings, organizations can foster a positive sense of commi...
Researchers at Chungnam National University have developed an AI model that uses deep learning to predict stable defect configurations in materials. The model, trained on data generated by conventional simulations, can generate results in milliseconds rather than hours, accelerating the material design process.
The Institute of Science and Technology Austria (ISTA) has received a significant donation to advance trustworthy AI technology. The €5 million gift from Garrett Camp will support fundamental research in artificial intelligence, focusing on interdisciplinary collaboration and long-term impact.
Researchers created a new frequency-aware approach to crafting adversarial images that better match human visual perception. The method, called Input-Frequency Adaptive Adversarial Perturbation (IFAP), significantly outperformed existing techniques in structural and textural similarity.
A large-scale study reveals that generative AI models have reached the threshold of average human creativity, but the most creative individuals still outperform even the best AI systems. The study also highlights the importance of human guidance and parameterization in modulating AI creativity.
A new AI model, ShapKAN, has been integrated into the cloud-based platform AI4Min-PE to predict five critical thermodynamic parameters across extreme pressures. The platform offers speed and accuracy far beyond conventional approaches, enabling scientists to explore the chemistry of materials under extreme conditions.
Engineers at the University of Pennsylvania have discovered that foams exhibit internal motion resembling deep learning in AI systems. The study suggests a common mathematical principle underlying both foams and AI training, with implications for designing adaptive materials and understanding biological structures.
A deep learning model trained on stage II colorectal cancer whole slide images accurately identified features linked to recurrence risk. The study found the model surpassed clinical prognostic parameters in predicting patient outcomes.
Researchers at Duke University have created a new method to use analog radio waves to boost energy-efficient edge AI, enabling devices to run powerful AI models without heavy chips or distant servers. The approach, called Wireless Smart Edge networks (WISE), achieves nearly 96% image classification accuracy while consuming significantl...
A new study found that popular AI tools for predicting river flow often misinterpret how heat and evaporation affect water, raising concerns for flood and drought planning. The researchers developed a hydrology-specific 'explainable AI' framework to uncover these issues.
Researchers at Mount Sinai have developed an AI-powered ECG analysis tool that shows promise in detecting Chronic Obstructive Pulmonary Disease (COPD) early. The model achieved high accuracy rates across diverse populations, including a subgroup with irregular heartbeat and smoking exposure.
Researchers developed G2PDeep, a web-based platform integrating six molecular data types to predict complex health outcomes. The platform enables better identification of omics-based molecular markers and improves personalized treatment strategies.
A new study found remarkable variation in how populations evolve in variable environments, with some cases benefiting from changes and others being hindered. The research has implications for understanding evolution and adapting to climate change, as well as informing AI and machine learning.
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.
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.
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 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 new HeiGIT dataset combines PlanetScope imagery with deep-learning models to analyze major transport routes, providing a high-accuracy global classification. The dataset supports better routing for logistics, infrastructure management, and emergency planning, highlighting disparities in road quality and its link to human development.
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.
A new AI model uses machine learning to predict drug toxicity in humans by identifying biological differences between cells, mice, and humans. The model improved predictive power over existing state-of-the-art models and demonstrated practicality in predicting market withdrawal due to toxicity.
A new security framework based on blockchain technology and distributed reinforcement learning ensures secure data storage and transmission while adapting to evolving threats. The framework demonstrated improved memory consumption and transaction latency compared to existing approaches.
The POINT platform integrates multiple biological networks, advanced algorithms, and a comprehensive biomedical knowledge graph to analyze drug-disease interactions. It combines node degree with deep learning methods to improve target prediction accuracy.
Autograph, a new framework, uses graph neural networks and deep reinforcement learning to achieve higher accuracy and faster execution of compute-intensive programs. It outperformed other approaches across various datasets, with notable improvements on Polybench, NPB, and SPEC 2006 benchmarks.
Researchers Prof Axel Cleeremans, Prof Anil Seth, and Prof Liad Mudrik warn that advances in AI and neurotechnology are outpacing our understanding of consciousness. They emphasize the need for theory-driven research and innovative methods to advance consciousness science.
Researchers warn that advances in AI and neurotechnology are outpacing our understanding of consciousness, with potential serious ethical consequences. A better understanding of consciousness could have major implications for AI, prenatal policy, animal welfare, medicine, mental health, law, and emerging neurotechnologies.
The Stowers Institute has appointed its first AI Fellow, Sumner Magruder, to harness the potential of artificial intelligence in biological research. He will collaborate with researchers to design new algorithms and unlock insights from large datasets.
Researchers develop AI-powered methods for modeling the Gulf of Mexico's dynamics, achieving higher accuracy for short-term predictions and emulating 10-year dynamics without hallucinations. This breakthrough drives forward critical management of natural resources in the U.S. and Mexico, advancing AI technology in earth sciences.
A novel machine learning framework combines interpretable deep learning with multiscale computational techniques to predict lattice thermal conductivity. The approach identifies high-performance materials for thermal management and energy conversion, providing deeper insights into heat transfer at the atomic scale.
Researchers at HUN-REN Szegedi Biológiai Kutatóközpont have developed an AI-powered platform for automated 3D cell culture analysis, enabling high-precision screening of cellular models. The technology removes the limitation of throughput in personalized medicine, allowing for fast and accurate analysis of clinical samples.
Recent research found that large language models are not yet able to consistently fool humans in conversations. They struggle with using discourse markers, opening and closing features, and subtle imitations. Despite rapid development, key differences between human and artificial conversations will likely remain.
Researchers at Edith Cowan University have developed a new way to measure biological age using AI, combining IgG N-glycome and transcriptome data. The method, called gtAge, predicts age with high accuracy and links to health markers, offering potential for early detection of age-related diseases.
VFF-Net applies label-wise noise labelling, cosine similarity-based contrastive loss, and layer grouping to improve image classification performance compared to conventional forward-forward networks. The algorithm reduces test errors on various datasets, enabling lighter and more brain-like training methods that make AI more sustainable.
MetaSeg achieves the same segmentation performance as U-Nets but requires 90% fewer parameters, making medical image segmentation more cost-effective. The new approach leverages implicit neural representations to quickly adjust to new images and decode accurate labels.
A new deep learning framework, Themeda, achieves high accuracy in predicting annual land cover categories across Australia's vast savanna biome. By integrating satellite data with environmental predictors, the model delivers probabilistic outputs that reflect uncertainty and captures ecological shifts at multiple spatial scales.
A novel framework integrates Kolmogorov–Arnold networks with dynamic predictor pruning optimization to improve TC intensity prediction. TCI–KAN achieves superior accuracy in 6-h intensity forecasts, outperforming referenced best records by 31%, 13%, and 6%. The model's accuracy varies by region and TC category.
Researchers developed an AI-based generative approach to discovering technology opportunities from patent maps using machine learning. The system translates patent vacancies into human-readable text, enabling the identification of untapped technologies and facilitating innovation forecasting.
A team of researchers developed a computational method that can design intrinsically disordered proteins with desired properties. The work uses automatic differentiation to optimize protein sequences and leverages molecular dynamics simulations for precision. This breakthrough has the potential to reveal new insights into diseases like...
A new 'future-guided' AI method developed at the University of California, Santa Cruz, has shown significant improvements in predicting seizures using brain wave data. The technique operates with two deep learning models working together, improving predictions further into the future by transferring knowledge.
A new deep learning approach, Electrode Net, accelerates the design of porous electrodes in electrochemical devices, achieving high accuracy and speed. The method outperforms traditional models on benchmarks, enabling rapid screening of large design spaces.
Biomedical engineers at Duke University developed a platform combining automated wet lab techniques and AI to design nanoparticles for drug delivery. The TuNa-AI platform resulted in a 42.9% increase in successful nanoparticle formation compared to standard approaches.
Researchers developed MoBluRF, a two-stage motion deblurring method for NeRFs, achieving high-quality 3D reconstructions from ordinary blurry videos. The framework outperforms state-of-the-art methods and is robust against varying degrees of blur, enabling smartphones to produce sharper and more immersive content.
A deep learning model achieved up to 98% accuracy in distinguishing autistic from neurotypical participants, providing clear insights into brain regions most influential to its decisions. The model could benefit autistic people and clinicians by offering accurate and explainable results to inform assessment and support.
Researchers developed an AI-driven algorithm that can predict nearly 70% of hot flashes before they're felt. The Embr Wave wearable device will incorporate this technology to mitigate symptoms and provide meaningful relief.
The book provides an overview of agent-based modeling and multi-agent systems, highlighting their application in understanding economic crises. It integrates machine learning to enhance adaptation and behavior of agents in dynamic environments.
A new machine learning model predicts heart disease risk in women by analyzing mammograms, offering a 'two-for-one' screening approach that combines breast and cardiovascular screenings. The model performs comparable accuracy to traditional risk calculators without requiring extensive clinical data.
A deep learning approach called Electrode Net optimizes porous-electrode design without sacrificing accuracy. The method achieves high predictive accuracy and speeds up computation time by 96%, enabling rapid screening of large design spaces.
Biochar, a carbon-rich material, is gaining attention for its ability to improve soils, clean water, and capture carbon. Machine learning models can predict biochar yield and pollutant removal efficiency with over 90% accuracy, accelerating its development.
A new tool called Flexynesis uses deep neural networks to evaluate multi-modal data, enabling doctors to make better diagnoses and develop more precise treatment strategies for patients. The tool is designed to be flexible and accessible to non-experts, bridging the gap in precision cancer therapy.
Researchers used AI models with integrated clinical and claims data to predict chronic kidney disease (CKD) progression to end-stage renal disease (ESRD). The study found that the models outperformed single data source models and reduced racial bias. The findings can inform likelihood and management of CKD, supporting targeted interven...
Researchers developed CerviPro, a multimodal deep learning model that accurately identifies high-risk patients with locally advanced disease. The model achieved superior predictive performance compared to conventional methods and provided critical prognostic insights.