Articles tagged with Artificial Intelligence
CURATE.AI is an AI-derived mechanism-independent technology platform that profiles individual patients based on their data and recommends optimal drug doses. This approach aims to improve treatment efficacy while reducing toxicity levels and addressing the challenges of personalized dosing.
Scientists at Osaka University created a flexible computing device from field-programmable gate arrays, increasing circuit density by 12 times and reducing energy usage by 80%. This innovation may lead to custom machine learning hardware with enhanced performance and reduced electricity consumption.
Researchers at Aarhus University have successfully automated the process of measuring and documenting terrain using drones, making it faster, cheaper, and easier. The AI system takes control of the drone's movement, adjusting its distance from the wall and camera position to capture high-quality images with minimal human intervention.
Researchers at Mount Sinai have developed an AI model that can accurately determine the type of lower back pain by analyzing clinical notes in electronic medical records. This breakthrough could lead to improved treatment options, reduced opioid prescriptions, and more accurate billing and therapy recommendations.
Researchers developed an automated AI technique to analyse CMR scan images, precisely quantifying blood flow to the heart muscle. The results showed a strong correlation between reduced blood flow and adverse health outcomes, including death, heart attack, stroke, and heart failure.
The selected scientists and engineers will participate in a week of intensive public engagement and science communication training, networking, and plan development. The fellows will return to their institutions with resources and connections to develop, implement, and evaluate their public engagement activities.
A 100-plus page report presents a vision for the strategic path to unleashing AI's full potential, with applications in healthcare, education, business innovation, and more. The roadmap aims to address the tremendous social change that will result from AI research.
A Tufts University team won a grand prize in the NSF Idea Machine Competition with their project 'From Thinking to Inventing,' focusing on what machines can invent and how. They aim to create resilient robots that can creatively solve problems, addressing pressing issues like climate change and wealth inequality.
Researchers at KAIST discovered a computational and neural mechanism for human meta reinforcement learning, enabling the porting of key human intelligence elements into AI algorithms. The study highlights how humans adapt to complexity and uncertainty in decision-making.
A new study uses an interactive voice application with AI to monitor patients' mental health, offering personalized analysis and finding accuracy comparable to physician assessments. The application, MyCoachConnect, analyzed patients' speech patterns and found unique changes over time, helping track mental states.
Researchers at the University of Warwick developed an AI technology that can detect hypoglycaemic events from raw ECG signals using wearable sensors with 82% reliability. This non-invasive method could replace fingerprick testing, particularly useful for pediatric patients.
Researchers at Universidad Complutense de Madrid developed an AI-powered algorithm that uses deep learning to detect irregularities in rice labelling. The algorithm achieved precision models between 93 and 99%, showing potential for future applications in the food industry.
The MIPT team developed a new method for reinforcement learning with demonstrations, enabling rapid solution of hierarchical problems in Minecraft. This approach opens opportunities for applying reinforcement learning to real-world tasks like autonomous vehicle control and manipulator control.
The Journal Healthcare Transformation: Artificial Intelligence, Automation, and Robotics explores the challenges and opportunities for integrating technology into consumer healthcare. A recent article found that five telemedicine companies provided nearly half of telemental health visits, highlighting the need for coordinated care.
The mind composes meanings from separate words through a dynamic interaction between brain regions, using multi-modal cues. Humans and artificial intelligence systems differ in representing meaning, with humans using contextual knowledge and emotional cues.
Researchers at the University of Waterloo developed an AI tool that uses deep-learning algorithms to determine if claims are supported by other posts and stories on the same subject. The system achieved 90% accuracy in stance detection, a key area of research in detecting fake news.
Researchers have developed an algorithm capable of identifying who in a social group would be the best persons to be trained as 'gatekeepers' capable of recognizing warning signs of suicide. The tool aims to improve efficiency of suicide prevention trainings for vulnerable populations, such as homeless youth.
Researchers at Stevens Institute of Technology have completed the first census of Wikipedia's 1,601 bots, shedding light on their functions and interactions with human users. The study found that bots play nine core roles, including fixers, connectors, and protectors, which together account for about 10% of all activity on the site.
The MIT researchers developed a bot equipped with artificial intelligence that can beat human players in tricky online multiplayer games where player roles and motives are kept secret. The bot uses deductive reasoning to identify friend or foe, ensuring victory over humans in certain games.
Researchers develop AI-driven platform to design, build, test and learn complex biochemical pathways for efficient production of lycopene, a red pigment used as food coloring. The system reduces time and cost by narrowing down possible pathways from over 10,000 to 100.
A recent study by NYU Abu Dhabi researchers found that bots posing as humans are more efficient at persuading partners to cooperate in a game. However, once their true nature is revealed, cooperation rates drop and their superiority is negated.
Researchers at Thayer School of Engineering at Dartmouth have developed an approach for detecting a speaker's intent to mislead, which could be used to extract opinion from 'fake news.' The framework uses universal features of deceptive reasoning and has been shown to outperform models based on verbal changes and personal differences.
A University of Texas at Austin team developed an AI system named Fashion++ that analyzes outfits and suggests tweaks for a more fashionable look. The tool uses visual recognition systems to consider color, pattern, texture, and shape, offering alternative outfits to users.
A new study published in INFORMS Journal Marketing Science found that AI chatbots can increase sales by four times compared to inexperienced workers. However, customer awareness of the use of AI chatbots before making a purchase leads to a significant decline in sales rates.
Samantha Kleinberg is developing artificial intelligence that provides personalized information to patients, enabling them to actively manage their health. The research focuses on using wearable data and machine learning to identify causes of diseases and develop more usable algorithms that consider human factors.
A multidisciplinary team from NYU Tandon School of Engineering is conducting foundational research to understand the impact of AI on expert work in healthcare. The project aims to redesign technology, workflow, and interactions to improve the effective use of healthcare technologies and human capabilities.
Researchers used a newly developed algorithm to study mixed x-ray images of the Ghent Altarpiece, separating features from the front and back of the painting's double-sided panels. The analysis improved our understanding of art masterpieces and provided new opportunities for art investigation, conservation, and presentation.
A new study aims to explore ways to share decision-making and information between humans and AI drivers, addressing concerns about perceived loss of control and fear of unknown actions. The goal is to increase user trust, consumer acceptance, and overall usability of autonomous vehicles.
An AI tool developed by USC computer science student Timothy Greer classifies song genres by analyzing lyrics and chords. The tool predicts 'Old Town Road' is country according to lyrics, rock according to chords, and pop when combining both. This method highlights the complexity of music perception and its relation to music construction.
A new study uses AI to predict food product recalls from Amazon reviews with 74-percent accuracy, identifying thousands of potentially unsafe products. The researchers trained an existing AI called BERT on customer reviews to identify recalled foods and found terms associated with FDA recalls in 20,000 other reviews.
Experts Eric Topol and Kai-Fu Li argue that international collaboration outweighs confrontation in the development of artificial intelligence for healthcare. They critique the forced divestiture of iCarbonX as misguided and highlight the need for access to clinical data to compete in AI health.
Researchers developed Pluribus AI that can play six-player no-limit Texas hold'em poker, outperforming top professionals and five copies of itself. The AI confirms traditional wisdom on limping as a suboptimal strategy for most players.
The National Science Foundation is funding a new supercomputer at the Pittsburgh Supercomputing Center (PSC) with a $10-million grant. The Bridges-2 system will provide massive computational capacity and flexibility to researchers in Pennsylvania and nationwide.
Scientists translate amino acid sequences into musical compositions, then use AI to generate new proteins with desired features. The method also enables the creation of bio-based musical pieces.
Recent Chinese AI research has made significant breakthroughs in big data analysis, with a focus on developing algorithms to detect abnormal data regions and repair historical correlations. Researchers have also proposed using Doppler measurements to improve target tracking performance in noisy environments.
Researchers developed an AI method to recognize patterns in infant cries, enabling healthcare workers and parents to accurately determine a baby's needs. The algorithm uses compressed sensing to process noisy environments, allowing for independent use across various scenarios.
The Radio Me project uses AI to adapt and personalize live radio, providing personalized reminders, information, and music for people with dementia. The system aims to improve quality of life and reduce hospital admissions by up to 30%.
Researchers found that when participants were unsure whether a profile was written by a human or an algorithm, they distrusted the one they believed to be machine-generated. The study suggests designing AI communication tools with transparency and trustworthiness in mind can improve user experience.
The Argonne National Laboratory's Aurora supercomputer will revolutionize scientific research and discovery with its exaFLOP performance and ability to handle both HPC and AI. The system is expected to have a significant impact on various fields, including cancer research, climate modeling, and veterans' health treatments.
The lack of regulatory standards in clinical artificial intelligence poses concerns about safety and efficacy, according to a recent study. The authors propose five standards to guide evaluation and regulation of clinical predictive analytics, emphasizing the importance of established clinical standards and clinician agency.
A new study examines individuals' willingness to adopt algorithms in decision-making contexts, revealing the importance of cognitive perceptions. Participants with low situational awareness were found to be more likely to delegate decisions to an AI algorithm.
A team of researchers from Ludwig-Maximilians-Universität München conducted a study on human perception and behavior towards robots. They found that when robots are humanized through anthropomorphic features, participants are less likely to sacrifice them to save human lives.
A new breed of algorithms has mastered Atari video games 10 times faster than state-of-the-art AI, using a combination of reinforcement learning and intrinsic motivation. This method autonomously identifies sub-goals and exhibits relatively human-like behavior, making it competitive with Google DeepMind in the game Montezuma's Revenge.
A novel model developed by MIT and Microsoft researchers identifies instances where autonomous systems learn from examples that don't match the real world, leading to dangerous errors. The model uses human feedback to pinpoint situations where more information is needed to act correctly.
Researchers explore how AI can analyze words and phrases, turn-taking patterns, and tone to improve medical communications. The technology has potential to revolutionize patient care by providing clinicians with assessments of their communication skills at lower costs.
The University of Texas at San Antonio has developed a framework using deep neural networks to classify and detect malware. This system was recognized as the top prize winner in the AICS 2019 Challenge sponsored by Crowdstrike Foundation and MIT Lincoln Laboratory, showcasing UTSA's expertise in cybersecurity.
A new software, idtracker.ai, can identify up to 150 individual fish with high accuracy, extracting valuable data for understanding group behavior. The AI-powered system uses deep learning neural networks and conventional algorithms to recognize unique features of each zebrafish.
A recent study by UC Berkeley reveals that current AI-powered tracking methods can identify individuals using their daily step data, posing significant threats to health data privacy. The research suggests that existing laws and regulations, such as HIPAA, are inadequate to protect individuals' confidential health information.
A UK-wide survey found that GPs are skeptical about AI's impact on their profession, with many believing technology can't replace empathic care. However, some GPs think AI can handle administrative tasks efficiently.
Researchers developed two classification models that gauge human trust in intelligent machines using electroencephalography and galvanic skin response. The models achieved a mean accuracy of 71.22 percent, enabling real-time trust estimation.
Researchers developed AlphaZero, a generalized game-playing program that can learn any game through self-play reinforcement learning. The system mastered complex games like chess, shogi and Go within hours of self-learning and outperformed state-of-the-art AI programs.
Cristiano Fanelli uses artificial intelligence to optimize particle detectors and study exotic hybrid mesons. He aims to apply deep learning techniques for accurate particle identification, revolutionizing nuclear physics research.
Researchers developed an AI Clinician system that predicts and recommends optimal treatment strategies for patients with sepsis, showing a higher chance of survival compared to human doctors. The system analyzed 100,000 hospital records and made more reliable treatment decisions than humans.
A new AI tool called DiffProf helps developers identify and optimize energy-draining features in their apps, improving battery life. The tool analyzes 'call trees' of similar tasks to show why certain features consume more energy than others, providing actionable diagnosis and fixes for developers.
Researchers at MUSC have developed a novel AI method that analyzes the neural networks of epilepsy patients to predict seizure-free outcome post-surgery. The technique, which maps the whole-brain connectome, achieved an accuracy rate of 79-88%, outperforming current clinical methods.
Researchers developed an ADC model that considers agent, deed, and consequence to explain variability in moral judgments. The study found that when stakes are high, consequences outweigh the nature of the deed, while lower stakes prioritize the agent's character.
The Quadratic Phenotypic Optimization Platform (QPOP) uses AI to analyze small datasets and design optimal drug combinations based on real experimental data. This approach can optimize and personalize combination therapy for individual patients with complex diseases, such as multiple myeloma.
Researchers at MIT CSAIL have developed a robot system that can inspect and understand objects, allowing it to perform specific tasks such as picking up objects from a cluttered space. The system, called Dense Object Nets, uses a self-supervised approach and doesn't require human annotations.
A commentary suggests using smart technology and artificial intelligence to improve the detection of sepsis in children in Canada. The authors propose automating data combinations of sepsis symptoms and other relevant information to create more accurate and low-cost trigger tools.
The National Institute of Biomedical Imaging and Bioengineering (NIBIB) at NIH will convene experts in academia, industry, and government to discuss the state-of-the-art of AI applications for medical imaging. The workshop aims to improve quality, reproducibility, and reliability of AI in medical imaging.