Articles tagged with Data Sets
Researchers developed a novel convolutional neural network for facial expression recognition, outperforming conventional models while being computationally less expensive. The new model achieved an accuracy of 72.4% using only 58,000 parameters.
A University of Washington study found that Pacific salmon in larger groups have lower predation risk, but may trade safety for food. The researchers analyzed historical data on fish catches and predator wounds to estimate group size and predation risk, revealing the benefits and costs of schooling in marine fish.
A newly expanded data set of brain scans from stroke patients called ATLAS now includes 1,271 MRI images with manually segmented lesions, facilitating large-scale stroke recovery research. Researchers hope to develop algorithms to automate lesion segmentation, enabling clinicians to predict patient responses to therapies.
A new study shows that community science activities can produce high-quality taxonomic data sets, empowering community scientists and supporting biodiversity discovery. Thousands of microscopic liverwort leaves were measured by community scientists over two years, with surprising accuracy.
A new study reveals that population declines have been greatest among species that migrate to areas with more human infrastructure, habitat degradation, and climate change. The research identified 16 human-induced threats to migratory birds, including infrastructure associated with bird disturbance and collisions.
Researchers have developed an app to help doctors identify patients with chronic lymphocytic leukemia (CLL) at risk of developing infections, allowing for earlier treatment. The app uses blood test results and genetic data to predict patient risk, improving treatment outcomes and reducing pressure on the healthcare system.
The American College of Cardiology and the American Heart Association have released a comprehensive data standard to standardize definitions for COVID-19-related cardiovascular conditions. This framework will help pool or compare data from various sources to assess its applicability to clinical practice and research endeavors.
A new report highlights the nation's race and ethnicity data shortcomings, recommending federal leadership on standardizing demographic data collection and reporting across critical issue areas. The report aims to help policymakers craft solutions that take into account where resources should be concentrated to address racial disparities.
Feminicide, a worldwide problem resulting in 87,000 intentional killings of women in 2017, has incomplete data due to undercounting and narrow legislation. Researchers using counterdata aim to enact alternative epistemological approaches centering care, memory, and justice.
A massive dataset from Binghamton University tracks public health government responses to COVID-19 at all levels of government worldwide. The dataset provides insights into the effectiveness of different types of government in responding to global crises.
A study by WVU researcher Ednilson Bernardes found that supplier pool pressure contributed to the oversupply of prescription opioids like oxycodone and hydrocodone. The researchers analyzed transactions from 2006-2012, revealing that over 90% of supply came from just three generics manufacturers.
A recent study published in Big Data & Society found that Democrats and Republicans in the US surveyed agree on the mathematical features of COVID-19 graphs, such as line slopes, but differ in their subjective interpretations. The researchers suggest that political affiliations may moderate how people interpret data, highlighting the i...
The INCLUDE Data Hub provides centralized access to large-scale research resources, including biospecimen libraries and clinical datasets, for the study of Down syndrome. With over 8,000 study participants and 30,000 biospecimens, researchers can accelerate discoveries that benefit people with Down syndrome.
The study collected sensor data and conducted surveys to understand the conditions people experience as they move about their daily life. The dataset provided a granular picture of temperature exposure, revealing that participants were well-situated to maintain low temperatures, but still faced dangerously hot environments.
Researchers developed a methodological framework to ensure web data validity in marketing studies. The study provides recommendations for addressing challenges in collecting and using web data via web scraping and APIs.
Researchers at MIT found that explanation methods used to aid human decision-makers in high-stakes situations often have lower accuracy for minoritized subgroups. The fidelity of these explanations varies dramatically between subgroups, with the quality often significantly lower for women and Black people.
A significant overestimation of arterial oxygen saturation by pulse oximetry was found in Asian, Black and Hispanic patients compared to non-Hispanic white patients. This discrepancy led to unrecognized and delayed recognition of COVID-19 treatment eligibility among these groups.
A WHO-supported series has collected 13,700 new database records on mosquito-borne diseases, providing valuable resource for studying and containing infectious diseases. The data can be used to train machine-learning models for vector detection and classification, improving global human health.
A Kyoto University study shows that even though biodiversity increases with more interactions, mean interaction strength decreases. The interaction capacity hypothesis proposes that this weakening of interspecific interactions may be due to increased community diversity.
A new tool called DeepSqueak uses deep learning to identify marine mammal calls with high accuracy, even in noisy environments. The tool was originally developed for rodent ultrasound signals but has been adapted to detect sounds at other frequencies, including humpback whales and delphinids.
A new analysis found that citizen science data from apps like eBird can be a valuable resource for researchers and managers, offering insights into wild bird populations. By comparing publicly-produced data with officially tracked numbers, researchers have identified patterns and potential areas for improvement.
Scientists have developed a machine learning algorithm that can accurately predict the lifetimes of different battery chemistries using as little as a single cycle of experimental data. The technique could reduce costs and accelerate the development of new battery materials, enabling researchers to quickly evaluate and test multiple ma...
A new study published in Animal Behaviour found that newborn African savannah elephants can keep up with their mothers' daily movements, contrary to previous assumptions. This remarkable ability allows the calves to benefit from protection against predators and integration into the herd's social structure.
A new study found that valbenazine, a VMAT2 inhibitor, is safe and effective in treating chorea in patients with Huntington's disease. Chorea is a common symptom of the disease, causing involuntary and irregular movements.
A recent study by the University of Rochester found that mobility patterns can be predicted with surprising accuracy based on data collected from acquaintances, even if individual users turn off their own location tracking. The researchers discovered that up to 95% of an individual's movement pattern can be inferred from people they ar...
A recent study combining climate data with fossil records of large mammals in Africa found that times of erratic climate change do not lead to major evolutionary changes. The research, published in the Proceedings of the National Academy of Sciences, suggests that environmental variability and species turnover may not be closely related.
Researchers at NYU Tandon School of Engineering propose a paradigm to solve the problem of inferring collective size from individual behaviors. By observing self-propelled Vicsek particles, they show that the time rate of growth of mean square heading is sufficient to predict the number of particles under particular parameters.
A new study by University of Warwick researchers finds that the adoption of cereal crops is the key factor in the emergence of complex hierarchies and states. Contrary to conventional theory, high land productivity does not lead to the development of tax-levying states.
Researchers develop a new method to reconstruct the states of complex nonlinear systems based on time series data. The approach optimizes reconstruction parameters by focusing on the geometric structure of attractors, resulting in improved accuracy.
A machine learning model has been trained to accurately identify individuals with post-traumatic stress disorder (PTSD) by analyzing text data. The model achieved an 80% accuracy rate in distinguishing between those with and without PTSD. This breakthrough could lead to the development of a cost-effective screening tool for health prof...
The BrainChart platform benchmarks brain development based on MRI data from over 100,000 individuals, creating a standardized chart like those for height and weight. The tool identifies previously unreported neurodevelopmental milestones and provides a common language to understand brain images from different sources.
Researchers created brain charts spanning human lifespan, revealing rapid brain growth in early life and slow decline with age. The charts use aggregated MRI datasets, covering a range of ages from fetus to 100-year-old adult, and aim to create a common language for describing brain development and maturation.
Researchers found that ocean reanalysis data sets exhibit differing evolutions during El Niño events, particularly as they develop and decay. The disparity among datasets grows with the event's peak phase and lingers even after neutralization.
A crowdsourcing campaign has compiled data on the drivers of tropical forest loss between 2008 and 2019, resulting in a high-resolution dataset. The analysis found that agriculture expansion, road construction, and wood extraction are major contributors to deforestation.
Researchers from North Carolina State University have released a dataset providing detailed information on COVID-19 vaccine shipments and wastage across the US. The data aims to inform modeling and decision-making to optimize vaccine supply chains.
Researchers analyzed health records of nearly half a million patients post-heart attack and found that depression was associated with a nearly 50% higher stroke risk compared to those without depression. The study highlights the need for greater attention to mental health in research and practice.
A new study using subsurface imaging sheds light on the geological connection between Yellowstone's iconic hydrothermal features and deeper heat sources. The research team detected hydrothermal alteration and found a remarkable similarity in deep structure beneath areas such as Norris Geyser Basin and Lower Geyser Basin.
A study has shown that wind variations over the southern Red Sea are the main drivers of sea-level extremes, driving levels up and down depending on wind direction. This understanding is crucial for coastal planning and management to mitigate the impact of storm surges and coastal erosion.
Researchers warn of machine learning bias when data published for one task is used to train algorithms for a different one. This can lead to compromised integrity and 'overly optimistic' results in medical imaging applications.
A new visual leaf library, developed by a Penn State-led team, provides a resource to help scientists recognize and classify plant leaves. The library contains 30,252 high-resolution images of cleared and fossilized leaves, allowing for rapid searching and comparison.
Researchers at Kaunas University of Technology improved an algorithm to detect Alzheimer's disease from MRI images, achieving over 98% accuracy. The new model uses a modified neural network and adapts to variations in data, such as differences in hospital equipment and patient positions.
Researchers from the Netherlands Institute for Neuroscience have published a dataset of electrophysiology data recorded from two monkeys' visual cortex during resting state. The dataset provides high-density receptive field coverage and can be used to yield new insights into background activity influencing visual information processing.
The collaboration aims to accelerate addressing unmet needs in the neonatal population by applying quantitative modeling, biomarkers, and regulatory science. Patients and families will benefit from improved understanding of bronchopulmonary dysplasia and better patient enrollment for trials.
Researchers at Texas A&M University developed a novel error estimator using transfer learning principles to evaluate machine-learning model performance. The technique enables fast screening of source data sets, improving the accuracy of diagnoses in complex medical issues like schizophrenia.
A new mathematical framework has been created to study fitness landscapes of regulatory DNA, enabling the prediction of gene expression changes. The framework uses a neural network model trained on millions of experimental measurements to decipher the evolutionary past and future of non-coding sequences.
Researchers at Duke University developed EyeSyn, a virtual eyes system that simulates human eye movement to train metaverse platforms. The system reduces privacy concerns and allows smaller companies to access the metaverse with minimal resources.
Researchers analyzed 280 confirmed sea turtle entanglements in Massachusetts waters and found that quickly reporting incidents can improve survival rates. The study highlights the importance of complete disentanglement to minimize injury and promote survival, with many turtles being alive weeks to years after being disentangled.
Researchers created a comprehensive genomic regulatory map of a 24-hour-old zebrafish embryo, identifying millions of regulatory segments that control gene transcription. The study used single-cell technologies and machine learning algorithms to analyze genome data from over 23,000 nuclei.
Researchers developed a machine-learning technique that can pinpoint anomalies in large datasets, such as power grid failures and traffic bottlenecks. The model uses advanced probability distributions to identify low-density values, allowing for faster and more accurate anomaly detection.
A new analysis of human remains provides the earliest DNA from sub-Saharan Africa, outlining major demographic shifts between 80,000 and 20,000 years ago. The study reveals people moved and settled in other areas, developed alliances and networks to trade and share information.
GIST researchers propose a new strategy for crime prevention using artificial intelligence, trained on a large-scale dataset of deviant incident reports and corresponding images. The model, called DevianceNet, can accurately classify and detect deviant places, making it a useful tool in urban safety development.
Researchers used AI to predict flood damage in the US, finding a high probability of flood damage for more than 1.01 million square miles across the country. The study suggests that recent FEMA maps do not capture the full extent of flood risk, with 84.5% of reported damage not within high-risk flood areas.
Researchers studied how diverse neural network training datasets impact generalization. They found that data diversity is key to overcoming bias, but also degrade performance when neural networks are trained for multiple tasks simultaneously. The study highlights the importance of designing diverse and controlled datasets in machine le...
A review by Nathan et al. showcases big-data revolution in movement ecology, revealing new insights into animal behavior and habitat use. Reverse-GPS systems, such as ATLAS, track animals with high accuracy, while acoustic telemetry tracks aquatic life, providing crucial data for conservation.
A UCI team uncovered key brain mechanisms by which the hippocampus organizes memories into sequences, enabling decision-making. The finding may help understand memory failures in Alzheimer's disease and other forms of dementia.
A new dataset from NYU Tandon School of Engineering and Woven Planet Holdings promises to help visually impaired pedestrians and autonomous vehicles navigate complex urban settings. The robust dataset uses over 200,000 outdoor images to test visual place recognition technologies that can improve navigation accuracy.
Researchers from Argonne National Laboratory have created a set of new practices to guide the curation of high energy physics datasets, making them more FAIR and reusable. The goal is to automate the finding and use of data for humans and streamline the development of AI tools for scientific discovery.
KAUST researchers have developed a flexible statistical model to analyze environmental data, revealing niches where existing methods fall short. The study provides a new approach to modeling dependence structures, highlighting the importance of understanding model limitations and extrapolation beyond observed data.
Researchers from Universidad Carlos III de Madrid developed an open-source development kit (PDK) to create solutions for businesses using personal data. The tool allows companies to exploit user data in a respectful way, promoting individual control over their data.
Researchers at NYU Abu Dhabi have published a comprehensive review of 50 fundamental traffic models using an extensive data set of 2.3 billion vehicle observations from 25 cities worldwide. The study found that a non-parametric model outperformed other traffic flow models, regardless of road type and congestion level.