Research applying artificial intelligence across science and engineering. Covers machine learning, sensing and quality control, immune and health prediction, and 5G networks.
Researchers at KAIST developed CURE, a technology that enables small AI models on smartphones to work efficiently with large models on servers, reducing server calls by an average of 55.61% while maintaining high accuracy. This approach allows for faster decisions with less server support.
The journal refines its evaluation criteria to emphasize translational impact, external validation, and actionable implementation in healthcare environments. Submissions will be evaluated based on real-world clinical applicability and systematic implementation of informatics solutions.
GrantsMate, an AI-driven research support platform, streamlines research workflows by integrating funding discovery, collaborator identification, and institutional policy guidance. The platform provides personalized recommendations and a conversational interface to simplify research administration and improve funding prospects.
SafeSeal embeds identifiable marks in LLM outputs while preserving content quality and robustness against removal attempts. It achieves this through named entity recognition and context-aware synonym replacement, ensuring high semantic similarity and entity preservation.
The Cross-Cultural Leadership Story Puzzle Game is an interactive, role-based platform that simulates authentic workplace challenges, encouraging cross-cultural communication and collaboration. AI-supported assessment provides personalized feedback, facilitating knowledge retention and improvement of leadership competencies.
The AgentHeart platform provides continuous cardiovascular monitoring, seamless integration into existing workflows, and real-time actionable guidance. It reduces diagnostic delays, improves clinical trust, and supports continuous clinical reasoning, ultimately aiming to transform cardiac healthcare practices.
The AI-driven system evaluates patent originality, marketability, and prior art analysis through integrated data processing and semantic analysis technologies. It incorporates third-party APIs and a Product-Market Fit scoring engine to assess commercial viability and forecast market relevance.
MyQL is an AI-powered tutoring system that integrates course material, interactive learning, and experiential learning to foster practical, hands-on learning. The platform guides students step-by-step through material, encouraging critical thinking and collaboration with AI.
Researchers at Harvard and Georgia Tech have developed RLE-Bench, a benchmark that tests AI coding agents' ability to engineer physical robots. The benchmark contains 48 tasks that test the agent's ability to perform engineering work required to build and operate robotic systems, including control and perception algorithms, designing r...
Researchers are using AI to model complex turbulent flows, including wind, aerosols, and combustion, to quantify uncertainty in foundation models. The project aims to create a physics-constrained AI foundation model for complex flows important to energy systems.
The study found that AI adoption has increased, but unevenly, with knowledge-intensive occupations and higher-educated employees benefiting most. However, there is a risk that existing inequalities will become further entrenched without targeted support measures.
Researchers at Karlsruhe Institute of Technology (KIT) have developed new ways to address the growing demand for fast and efficient data exchange. They have created photonic microchips and optical bridges that enable the economical production of optical systems for data networks, data centers, and artificial intelligence.
Researchers developed a multitask deep learning framework to predict how strongly a material adsorbs sulfur gases and how effectively it senses them. The approach accelerated the discovery of materials for gas detection and purification, highlighting specific material candidates with strong sensing responses to toxic gases.
A recent study published in iScience found that AI-powered machines have difficulty recognizing objects from their overall shapes when aspects of an image are distorted. Humans, on the other hand, are able to leverage the global shape cue for visual object recognition, a skill that current AI models do not replicate.
A new AI-powered test, IICM+, has been developed to predict breast cancer recurrence risk more accurately than the widely used 21-gene Recurrence Score. IICM+ uses clinical, molecular, and histopathology data to provide reliable prognostic information, distinguishing between patients at different risks of recurrence.
The summit explores the intersection of AI, eye care, and public health, with a focus on artificial intelligence, oculomics, population health, and myopia management. The event brings together leaders from healthcare, technology, and optometry to discuss innovative approaches to improve eye health and expand access to care.
Researchers at the University of Bristol have developed a new AI method inspired by the game of 20 Questions, which reduces training costs and complexity by combining simple yes/no questions. This approach can perform complex classification tasks at a lower computational cost, making it more suitable for real-world use, particularly in...
INSEAD's immersive AI cases use AI to create experiential learning experiences, combining human insight with AI-driven personalized feedback. The platform now offers over 40 AI-powered learning experiences, transforming business education and preparing leaders for an AI-enabled world.
A virtual biotech company powered by AI agents has made groundbreaking discoveries in drug discovery, including a biological signal that predicts which drug candidates are more likely to succeed. The company has also designed a cancer therapy that was later independently built by a major pharmaceutical company.
By creating virtual cells from 4D AI models and digital twins, researchers can predict how mitochondria respond to drug treatment, grouping cells that respond similarly together. This technology has the potential to speed up drug discovery and accelerate research for diseases such as cancer, diabetes, and Alzheimer's.
Researchers used AI to screen a drug library for Streptococcus pneumoniae, identifying 11 compounds with antibacterial properties. Nine of these compounds inhibited the growth of the bacteria, including one that showed efficacy against drug-resistant strains.
A new AI approach helps distinguish genuine SSD failures from false failure reports in large-scale data centers, improving reliability and efficiency. The model achieved an F1 score of 0.717 under a 40% false-failure rate, outperforming conventional models.
A Stanford Medicine team has developed an AI program called Paper2Agent that turns scientific manuscripts into interactive agents, enabling chat and collaboration between agents. This innovation has the potential to accelerate research and discoveries, particularly in areas like genetic mutations and ADHD risk.
Researchers developed xvr, a patient-specific AI technique that accurately matches X-rays with 3D medical scans, improving surgical navigation and safety. This innovation enables faster and more precise minimally invasive surgeries, particularly in fields like orthopedics and neurosurgery.
Researchers identify recurring states of tumor microenvironment that predict response to immunotherapy, finding that early treatment-induced changes can anticipate course of immune response
Researchers used human toddlers' vocalizations to investigate AI's limitations in deciphering animal communication, finding that AI models failed to classify vocalizations according to their meaning. The study suggests that combining AI tools with behavioral observations and other methods is needed to truly decipher animal communication.
Researchers replicate 2,000-year-old burned papyrus scrolls, test a new method to decipher Herculaneum texts using X-ray tomography and leaded ink detection. The study provides a simple way to identify scrolls with leaded ink, which can be virtually unrolled and read.
Researchers call for health-literate AI to design and govern AI-mediated health communication, ensuring people can understand what matters, make informed decisions, and know what to do next. The framework proposes four principles: comprehension, agency, accountability, and proportionality.
Scientists systematically map the Biginelli reaction to uncover a previously unknown branch that produces complex bicyclic structures and molecules with unusual supramolecular behavior
A University of Toronto study found AI scribes can reduce the time physicians spend documenting patient visits by 69.1% during simulated primary care appointments. The technology can also lead to more timely documentation and reduce cognitive labour for physicians.
Marcos Zampieri is developing a framework to teach students about AI-assisted software development. The project aims to improve student learning outcomes and promote responsible AI use in introductory computer science education.
The conference will explore next-generation biologics and immunotherapies, targeted protein modulation, and emerging therapies for oncogenic drivers. Experts will discuss the role of artificial intelligence in accelerating therapeutic development.
The rapid adoption of AI is creating new opportunities to improve open source software, but also placing growing pressure on maintainers. AI can help identify vulnerabilities and develop patches, but its capabilities can also be used by attackers.
Dresden researchers create an on-premise medical AI system that supports diagnoses and clinical decision-making, while protecting sensitive patient data and enabling clinicians to assess AI-generated result reliability. The system achieves high diagnostic accuracy in standardized tests, with consistent answers indicating correct diagno...
Researchers developed a statistical SERS strategy to turn signal fluctuations into concentration fingerprints, enabling more reliable ultrasensitive quantitation. By analyzing the full continuous SERS intensity distribution, the team achieved 100% identification accuracy across diverse chemical and biological applications.
Researchers at Eindhoven University of Technology have developed a new chip that enables AI computations closer to the user, reducing energy consumption and reliance on data centers. By moving AI computations to the edge, data does not have to travel back and forth to a data center, making applications faster and more privacy-friendly.
A study by Penn State researchers found that conversational chatbots can lead users to trust misinformation, but adding verification tools can offset this effect. The study tested the impact of conversational AI chatbots on user trust and found that users who could verify information were more skeptical of the AI's responses.
Yu Meng's research focuses on weak supervision, allowing AI systems to learn from incomplete, noisy, or inconsistently labeled data. His methods could improve AI in fields like healthcare, information retrieval, and scientific discovery.
The study demonstrates the FIND Lp(a) model's ability to identify individuals with high Lp(a) more than twice as likely as the overall population with ASCVD. The model supports targeted Lp(a) screening, accelerating universal screening adoption and enhancing cardiovascular risk management.
Graphene diffractive zone plates can produce wavelength-dependent focal and interference patterns that serve as physical responses. AI analysis transforms these patterns into compact binary security responses.
The SUNY Technology Accelerator Fund is providing grants to support research in AI, energy-efficient semiconductors, and non-invasive monitoring. Five campuses will receive funding for projects that could improve cancer diagnostics, circadian rhythm disorders, and burn diagnostics.
MIT researchers developed a new technique to help generative AI models meet strict safety requirements without sacrificing output quality. The 'HardFlow' algorithm reformulates hard-constrained sampling as a trajectory-optimization problem, enabling subtle corrections while enforcing hard constraints.
Rocky Scopelliti argues that AI systems' growing self-awareness and moral dispositions require immediate practical ethical concerns to be addressed. He suggests protocols for AI systems, especially those trained to care, to understand context and make moral judgments.
Researchers at Stanford Medicine have developed AI cell models, called universal cell embedding and TranscriptFormer, to compare cells among species and gain insights into diseases. These models can identify cell types, evolutionary relationships, and even distinguish between healthy and diseased cells.
Researchers will develop AI-powered tools to examine cascading wildfire impacts, including flash floods, debris flows, and infrastructure disruptions. The project aims to bolster public safety and community resilience by combining AI, infrastructure modeling, and social science.
A new study led by Penn State researchers compared AI chatbots' moral decision-making with that of humans in a hypothetical kidney transplant scenario. AI models often fixate on single factors, diverging from human values, and confidently pick a decision despite uncertainty. In contrast, humans frequently express indecision due to a de...
The commission, composed of 30 global experts, aims to assess and strengthen the value of academic institutions to society, build public trust, and support their contributions to society. The group will analyze foundation and implications of academic responsibility and provide actionable recommendations.
Researchers propose a framework for packaging that senses, learns, and acts to reduce food waste and spoilage. The system uses AI to interpret signals from sensors embedded in the packaging, enabling real-time monitoring and adaptive responses to minimize waste and optimize food distribution.
Researchers explore the potential of shadow AI in healthcare, voice analysis for early disease detection, and digital resurrection technologies for bereavement. A recent FDA approval also highlights the therapeutic benefits of video games for ADHD treatment, with potential for personalization and immersion in the future.
A new review in Science Bulletin outlines how multimodal AI can integrate molecular, cellular, and clinical data to refine treatment choices and predict side effects. The approach aims to address challenges in cancer treatment, including primary resistance and immune-related adverse events.
Researchers developed a graph-based approach to directly extract concise and accurate constitutive equations from solid material experimental data. The method outperforms mainstream empirical models in predictive accuracy while preserving explicit, physically interpretable mathematical formulations.
JMIR Publications and ZB MED extend their Flat-Fee Unlimited Open Access Publishing Agreement for two years, covering over 30 Gold Open Access journals with zero Article Processing Charges (APCs) for participating German research institutions. The new agreement provides predictable and sustainable funding for open access publishing.
Researchers developed an AI-based system to detect temperature stress in fish, revealing diverse temperature tolerance among Medaka fish and closely related species. The system accurately predicts the effects of climate change on fish, with implications for conservation and large-scale comparisons among strains and species.
The grant will support research on AI tools that can generate interactive narratives effective at engaging K-12 students and improving learning. Researchers will develop self-improving systems, emerging narrative generation technologies, and agency-centered learning approaches.
A SUNY Cortland faculty member is part of a research team exploring how young children engage with AI and addressing equity challenges in early childhood education. The project aims to create a four-year research agenda focused on sustainable solutions that can scale nationally.
A new model forecasts subway passenger destinations and travel times one day ahead by analyzing multi-time-scale patterns in rider behavior. The model outperforms widely used forecasting methods, enabling subway operators to better plan train schedules and manage congestion.
Generative AI is changing the development pathway for software developers, reducing opportunities for hands-on experience and trial and error. The study found that senior developers are increasingly using AI to handle tasks previously assigned to junior developers, making it difficult for juniors to gain expertise.
Researchers found that AI-generated images can improve species identification and biodiversity monitoring when real images are limited. However, the synthetic images were less effective than real images overall, highlighting the importance of community science and real-world observations.
A new scoring system developed by researchers at Upstate Medical University can help identify patients at high risk of intramyocardial hemorrhage after a heart attack, guiding treatment and monitoring. The system uses explainable artificial intelligence to predict patient risk in real-time, allowing doctors to make informed decisions.
A new learning mechanism uses natural variability in neural activity to understand how synapses adapt and improve the learning capabilities of brain-inspired devices. The mechanism, called Spike-based Alignment Learning, solves the weight transport problem and matches the performance of existing approaches without unrealistic assumptions.