Research applying artificial intelligence across science and engineering. Covers machine learning, sensing and quality control, immune and health prediction, and 5G networks.
A telemedicine-based management programme improved quality of life and self-care in patients with heart failure, with clinically relevant improvements in health status and symptoms. The programme demonstrated high patient satisfaction and adherence to daily habits, such as medication intake and physical activity.
A study found that having an older AI-estimated heart age compared to actual age is associated with worse cardiovascular and dementia-related death outcomes. The research used deep learning AI models to estimate heart age from ultrasound images and analyzed data from a community-based cohort of 568 adults. The results showed a signific...
A team of PSU researchers aims to develop evaluation tools for AI's response to students' mathematical reasoning, focusing on feedback quality and relevance. The two-year project will use real math tasks and data visualization reasoning to test AI models' ability to recognize and respond to student thinking.
The authors propose materials world models to maintain a persistent representation of material states, histories, and observations. The models connect AI algorithms, multiscale computation, and physical experimentation to ensure scientific continuity across material evolution and evidence interpretation.
The NTU-Canaan Green Intelligent Compute Collaboration aims to develop energy-aware algorithms and privacy-preserving edge systems for AI applications. The four-year initiative addresses the growing demand for computing resources and the need for sustainable computing systems.
ISTA PhD students Riccardo Cadei, Jiale Chen, and Nikita Kalinin have been awarded Google PhD Fellowships for their groundbreaking AI research. Cadei and Chen will receive up to $120,000 in funding to explore cause-and-effect relationships and develop more efficient large language models, respectively.
Researchers introduced a framework for evaluating AI tools in psychological assessment, finding that AI models can score well on accuracy measures while missing key signs of mental health concerns. The study recommends a more rigorous testing process to ensure valid psychological assessments.
This book provides a structured approach to evaluating AI trustworthiness in finance, considering explainability, causality, and bias indication. It advances a multidimensional notion of trustworthiness, requiring concrete operational mechanisms beyond high-level principles.
A Saitama University research team developed an architecture called layered attention-enhanced reservoir computing (LAERC) to achieve competitive predictive performance in language modeling while reducing computational burden. The study demonstrates that reservoir-based architectures can provide a practical alternative to traditional t...
Researchers at Dartmouth College found that pink noise, a conversational recipe balancing familiarity and unexpectedness, tends to be rated as more enjoyable. The team's study suggests that injecting randomness and switching topics can help revitalize conversations and improve human communication.
A new AI system, PI-GAE, predicts aerodynamic performance of shapes, simplifying the design process for engineers. The system accurately reconstructs shapes and predicts performance, with coefficients of determination ranging from 0.946 to 0.992 across four tests.
Neuromorphic memristors mimic biological neurons and synapses, enabling artificial synaptic plasticity and neuron-like behavior. These devices offer opportunities to improve computational efficiency and reduce power consumption.
Researchers have found evidence of decentralized mass production of papers based on secondary analysis of public health data. The papers are often based on data from the Global Burden of Disease project and are produced using commercial software packages sold by Chinese companies.
Mayo Clinic researchers created an AI model that predicted kidney stone composition with 73% accuracy, compared to 33% with conventional calculations. The model analyzed 24-hour urine test results to identify patterns and predict stone type, offering a new source of clinical information for healthcare professionals.
Researchers developed a blood test that distinguishes between women with early-stage breast cancer and healthy women with 95% accuracy. The test also predicts the risk of recurrence after treatment, detecting aggressive cells with distinct epigenetic signatures.
A new low-cost test method developed by University of Illinois engineers links cement particle shape to construction costs, promising cost savings for the industry. The method uses a $500 microscope-based framework to measure both particle size and shape, which affects cement behavior when mixed with water.
MIT scientists developed a training method to generate models that simulate climate effects much more quickly than other approaches, allowing for more reliable estimates and unexpected outcomes identification. The technique boosts performance and adaptability while using only a fraction of the data required, significantly cutting produ...
The EU's proposed Cloud and AI Development Act (CADA) aims to reduce dependence on non-European providers, but researchers argue it's not ambitious enough. The proposal focuses too heavily on expanding data centre capacity and doesn't adequately address sustainability and social impact.
A recent study reveals that AI can contribute to a more sustainable construction industry by optimizing energy use, improving resource efficiency, and enhancing climate resilience. However, the technology is not a standalone solution and has its own environmental footprint, which must be considered when evaluating its climate benefits.
Researchers at Graz University of Technology developed CLARA, an AI toolbox that accelerates bioreactor simulations by reducing complexity while maintaining physical accuracy. The tool enables precise control of large-scale industrial plants, optimizing processes in fermentation, biomass energy, and polymer production.
A new method, GlioTrace, reveals how individual glioblastoma cells spread through the brain and adapt to treatment, offering insights into tumour cell invasion and potential therapies. The study also found that cancer treatments can affect tumour cell movement and shape, and that some cells may be more treatment-resistant than others.
Two PolyU research projects, Smart Motion Capture App for Rehabilitation and TruTru, have won CES 2027 Innovation Awards in Digital Health and Artificial Intelligence categories respectively. The apps utilize AI and augmented reality technologies to provide personalized rehabilitation and emotional well-being support, respectively.
The 14th Shanghai Forum on Lifelong Education highlighted the need for universities to rethink their role in serving society, leveraging international experience while staying grounded in Chinese practice. Key takeaways include the importance of expanding lifelong learning opportunities, strengthening collaboration with lifelong educat...
The Heidelberg Laureate Forum offers 200 exclusive spots for young researchers to network and exchange ideas with laureates of prestigious prizes in mathematics and computer science. Applications are now open until February 11, 2027.
A new system uses AI-powered smart glasses and sensors to assess vocational skills, providing fairer and more scalable evaluation of practical skills. The system uses computer vision and machine learning to score skills, with expert adjudication for human-centered skills.
A Northwestern University study analyzing 5 million public comments on pollution policy found that economic concerns carried more weight than climate and public health benefits. The research highlights the complex process of shaping government policy and the potential for competing interests to slow progress toward cleaner transportation.
A study of 10,000 employees found that sit-stand desks increase active computer time by 43 minutes a day, resulting in 171 additional hours of work per year. This translates to a potential annual productivity gain of nearly $9,000 per employee.
RIKEN-AXIS aims to advance co-creation in AI for Science by integrating cutting-edge infrastructure and research. The initiative enables humans and AI to collaborate to expand scientific exploration, leveraging agentic AI and open scientific research platforms.
Researchers at the University of Waterloo have developed a hybrid cooling system that targets hotspots on computer chips with a single drop of boiling coolant, reducing temperature by up to 17°C in 15 seconds. The system combines immersion cooling with targeted spot cooling, offering a potential solution for thermal management in data ...
An experimental study found that AI can enhance creative tasks, but human involvement is crucial for quality and satisfaction. The 'Human Creativity' and 'Copilot' conditions showed higher quality and satisfaction rates compared to 'Human Confirmation' and human-only approaches.
A newly developed AI agent called MLP Copilot has made atomic modeling easier by bridging computational tools and scientific evidence with human supervision. The framework has been used to investigate aluminum nanoparticle oxidation, revealing two behaviors of particles with an existing oxide shell.
Researchers used machine learning techniques to analyze thousands of species, finding that fish school independently 60 times and lose the trait 320 times over 350 million years. The study provides insights into the evolution of collective behavior, challenging previous assumptions about its rarity.
The New York Center for AI Responsibility and Research will bring together 80+ researchers to test and evaluate AI systems, addressing safety and security concerns. The Center will also develop scientific tools and standards for evaluating AI systems, supporting responsible AI use in various fields.
Researchers at KAIST developed GAZE, a filesystem technology that improves data management in next-generation flash storage devices, increasing data-processing performance. The technology reduces cleanup tasks and improves the system's efficiency, enabling faster data supply for AI inference and other high-performance applications.
A systematic review maps generative AI models and their applications in medicine, highlighting challenges and future directions. The review proposes developing uncertainty quantification and explainable AI to build trustworthy generative AI systems for medical professionals.
A new AI tool, PREVIA, analyzes patients' narratives about injuries to predict who will later develop PTSD, correctly identifying over 95% of at-risk patients. This tool could help clinicians catch the condition early, as it is designed to be a quick and efficient screening method.
Researchers have developed a new approach called agentic robotics, which enables teams of AI agents to generate, test, and refine robot programs quickly and efficiently. This method addresses the challenges of collecting and processing large amounts of data and programming robots, allowing for faster and more adaptable robot development.
Researchers at Lehigh University are developing a model that uses causal auditing and symbolic reasoning to improve the transparency and accountability of LLM-based recommender systems. This could help users better understand why a particular recommendation was made and what information influenced it.
Researchers have developed a unified AI architecture, WorldParticle, that can simulate multiple physical phenomena, from sliding sand to folding proteins, with one unified artificial intelligence architecture. The system represents every material as tiny particles and predicts where those particles will move one step at a time, adaptin...
Researchers used AI to predict cavity risk by analyzing social determinants, enabling targeted public health programs. Wealth, household density, and water source were strong predictors of cavity risk, with non-White populations and private dental insurance showing protective effects.
Researchers from Brown University map the sprawling digital ecosystem that enables AI-generated nonconsensual intimate images, highlighting the need for policymakers to address the 'ecosystem' to prevent abuse. The study provides a taxonomy of 11 technology categories and identifies a gap in current mitigations, which are mainly reacti...
A new study proposes a holistic theory of AI fairness and an applicable framework for organizations to manage the complicated issue of fairness in AI-automated decision making. The proposed design theory, FAIR, addresses conflicting concerns through regular human-AI collaboration and treating AI fairness as a continuously managed probl...
A 60-page review assesses the validity of AI-driven science, highlighting three levels of evidence and four constraints: data quality, model interpretability, generative physical hallucinations, and human oversight. The review suggests that the integration of data, compute, and people is key to achieving advantages in AI-for-science.
Researchers from the University of Johannesburg propose an AI-driven precision farming approach to improve indigenous livestock breeds' heat-resilience. By analyzing gut microbiome data, local AI models can optimize feed management and animal care, increasing production and reducing diseases in African farming operations.
AI is now a fundamental tool across 11 fields, driving discoveries and breakthroughs. From physics-informed neural networks to generative AI risk management, AI applications are being used to advance knowledge in various disciplines.
Researchers use AI to screen existing compounds for therapeutic potential, identifying novel compounds that bind to specific receptors. The framework, known as Full-Spectrum Generative Lead Discovery (FSGLD), uses generative AI to uncover patterns in molecular fingerprints that point to useful properties.
AlphaOracle cuts analysis time by 64% and 78.7% of experts rate its usefulness; the system uses four stages to trace characters, including morphological analysis and contextual alignment, and is presented as a step for partially documented scripts
Researchers propose connected research system linking living databases, AI simulations, and large language models to automate experiments and improve material discovery. The goal is a 'closed loop' where data, simulations, experiments, and researchers continuously learn from each other.
AWARE_AI, an innovative AI program, investigates how human experiences of healthcare are connected to wellbeing. The study found associations between positive experiences of integrated care and positive assessments of wellbeing.
Researchers designed a compact silicon modulator with integrated signal equalization to improve bandwidth and reduce reliance on complex digital processing. The device achieved data rates of 180 Gb/s and 200 Gb/s, offering potential for high-capacity, energy-efficient optical connections in future AI data centers.
Researchers developed an ultrafast all-optical biomimetic encoder for spiking neural networks, achieving high-throughput computation and precise fiber-compatible encoding. The system reaches 94.1% classification accuracy for handwritten digit recognition and demonstrates scalability for diverse real-life signals.
A new generation of AI models could help scientists understand ecosystems and predict their responses to environmental change. Ecological foundation models, a new approach to AI, could learn from diverse sources of information and apply that knowledge to a wide range of ecological questions.
The National MS Society is launching a funding opportunity to support innovative screening technologies for early MS detection. The goal is to accelerate progress in identifying individuals at risk of MS, ultimately reducing time to diagnosis and improving long-term outcomes.
Scientists from 25 countries shared advances in epidemic intelligence, pathogen genomics, and vaccine development to strengthen global health preparedness. The meeting reinforced the importance of international cooperation in addressing current and future health challenges.
A study by economists at University of Warwick and Imperial College London found that 5.1% of UK adults, mainly professionals and finance workers, paid for AI tools like ChatGPT or Claude. The market grew from almost nothing in early 2023 to a billion-pound annual market, with ChatGPT leading but Claude gaining ground rapidly.
Researchers review how AI enhances continuous glucose monitoring to enable predictive, personalized diabetes management, improving glycemic outcomes and prioritizing high-risk patients. Machine learning models forecast individualized responses to meals, enabling tailored dietary and physical activity guidance.
Researchers at the University of Tokyo developed a new AI framework that uses light to process information, reducing computational cost and energy consumption. The framework, combined with an optical computing system, achieved promising results and could pave the way for faster and more energy-efficient AI hardware.
Researchers at the University of Sydney have identified a fundamental mathematical reason why AI systems can learn different behaviors even when trained under almost identical conditions. The study found that extremely small differences in a model's starting conditions can lead to different outcomes, revealing a fractal structure in the
Experts from Hackensack Meridian Health and other institutions emphasize the need to make human-AI interactions visible to clinicians to advance patient care. The authors propose using divergence visibility to reconcile human and machine opinions, mitigating liability and improving clinical decision-making.
A University of Cambridge team has developed an AI-guided bird flu vaccine that provides complete protection against globally circulating variants in animal trials, outperforming current stockpiled vaccines that offer only 50% protection. The new vaccine uses a 'plug and play' approach to create safe, highly effective vaccines for a pl...