Articles tagged with Data Sets
Researchers from the University of Kansas have created a powerful dataset to facilitate drug development against gram-negative bacteria. The dataset reveals over 270,000 previously unidentified outer-membrane proteins with potential as vaccine targets.
Researchers developed a deep-learning model to assess CXR images for probable COVID-19 severity. The model achieved an area under the receiver operating characteristic curve of 0.78 when predicting intensive care need within 24 hours.
Biased AI can limit climate predictions and misguide governments due to missing information from under-represented communities. Human-in-the-loop design can fill these 'data holes' by offering a sense check on used data and context.
A new study by University of East Anglia reveals ChatGPT's systematic left-wing bias, favoring Democrats in the US and Labour Party in the UK. The platform's responses also lean towards President Lula da Silva of Brazil's Workers' Party.
Research identifies key molecular signatures and pathways contributing to skeletal muscle strength loss in females with estrogen deficiency. The study found parallel patterns of inhibition and activation across various signaling pathways, including AMPK and calcium signaling.
Researchers developed an innovative optical tool, the Schistoscope, to capture microscopy images of urine samples for efficient detection of Schistosoma haematobium eggs. A two-stage diagnostic framework using deep learning accurately identified and counted eggs in field settings with high sensitivity, specificity, and precision.
A team of researchers developed a novel method that leverages temporal characteristics of blood pulse to estimate heart rates with improved accuracy, especially in scenes with ambient light fluctuations. The proposed method showed a 36.5% improvement in estimation accuracy compared to conventional methods.
A Swansea University-led study found that Welsh breastfeeding rates increased during the pandemic, with a significant correlation between mothers' intention to breastfeed and exclusive breastfeeding duration. The study proposes targeted interventions during pregnancy and policies to support families to improve breastfeeding duration.
A sociological study by the University of Zurich confirms that many professionals consider their work to be socially useless. Office jobs were found to be more than twice as likely to feel pointlessness compared to other occupations. The study suggests that factors such as routine work, job autonomy, and management quality also contrib...
Researchers combined linguistics and genetics to propose a new origin theory for the Indo-European languages, suggesting an ultimate homeland south of the Caucasus. The study estimated the family to be approximately 8100 years old, with five main branches split off by around 7000 years ago.
A new study uses machine learning to analyze data from DrugAge, a database of chemical compounds modulating lifespan in model organisms. The researchers create four types of datasets to predict whether or not a compound extends the lifespan of C. elegans, using features such as compound-protein interactions and Gene Ontology terms.
Researchers have developed a computational tool to compare large datasets and predict immune responses to disease, potentially leading to better vaccines. The new algorithm, designed by La Jolla Institute for Immunology scientists, uses machine learning to identify underlying patterns in immune system data.
A new study suggests capping the energy use of the top 20% of consumers could reduce carbon emissions by 11.4%. The strategy would allow those with lower incomes to increase their consumption levels, promoting fairness and delivering climate justice.
A new AI technology has been developed to generate artificial scientific data, allowing for faster and more efficient detection of material features. The AI uses generative adversarial networks to incorporate background noise and experimental imperfections into the generated data, making it virtually indistinguishable from real data.
Researchers propose a novel vehicle color recognition method based on Smooth Modulation Neural Network with Multi-Scale Feature Fusion, achieving high accuracy and overcoming class imbalance issues. The proposed method outperforms state-of-the-art VCR methods and meets the requirements for fine classification of vehicle colors.
A recent study by Kyoto University has raised concerns about the authenticity of Big Oil's net-zero emissions claims. The research team found that oil majors are not making sufficient progress in phasing out fossil fuels and transitioning to clean energy, despite their pledges to achieve net-zero by 2050.
BioAutoMATED is an all-in-one AutoML platform designed for biologists, enabling easy analysis and interpretation of biological sequences. The platform uses three existing AutoML tools to generate models that can predict biological functions from sequence information.
A retrospective analysis of national data found that over 20 million Americans experienced loss of smell or taste after COVID infection, with a large portion never fully recovering these senses. The study estimated that almost 28 million Americans may be left with decreased sense of smell after COVID infection.
Researchers analyzed 692,534 race times to find genetic improvement accounts for 60% of speed increase in short-distance races, while heritability is low across all distances. The study suggests weaker selection or other factors limiting genetic progress, particularly over long distances.
Researchers identified five subtypes of heart failure using machine learning, including early onset and atrial fibrillation related. These subtypes have different mortality risks, with some patients at higher risk of dying within a year after diagnosis.
A team of scientists has developed an automated algorithm to reconstruct the shape of each neuron inside a light microscopy image using deep learning. This breakthrough addresses the challenge of generalizing algorithms across diverse species, brain locations, developmental stages, and microscopy image sets.
A deep learning-based framework called EMGSense enables accurate wearable EMG device usage through AI self-training techniques. It achieves average accuracy of 91.9% in gesture recognition and 81.2% in activity recognition, outperforming state-of-the-art approaches.
Researchers tracked immune cell clusters in the aging mouse prostate using highly multiplexed immune profiling. Early adulthood sees myeloid cells, while between 6-12 months old, there's a profound shift to T and B lymphocyte-dominance. The study reveals new insight into prostatic inflammaging and the window for interventions.
Researchers used a multiomics approach to analyze changes in transposable elements after influenza A virus infection, identifying transcription factors contributing to individual responses. The study provides insights into the variable severity of illness among individuals infected with the same virus.
A research group led by NCKU professor I-Non Chiu conducted the first cosmological study on galaxy clusters identified by eROSITA, analyzing 550 galaxy clusters. The results suggest that Dark Energy occupies up to 76% of the total energy density in the Universe.
Online radiologists choose studies based on financial attractiveness, leading to delays in high-priority cases. This study found that expedited priority class contained the highest percentage of delayed studies.
A new guide has been created to standardize fossil pollen datasets, enabling researchers to compile and analyze large-scale syntheses of palaeoecological data. The FOSSILPOL workflow and R-package provide a step-by-step process for handling data preparation, ensuring good data quality and minimizing erroneous interpretations.
Researchers found that machine-learning models trained with descriptive data label rule violations more harshly than humans, leading to potential serious implications in the real world. This study highlights the need for careful consideration of data labeling and training methods to ensure fairness and accuracy in AI decision-making.
Researchers used cancer proteomics data to identify gene candidates for therapeutic targeting, focusing on protein kinases in uterine endometrial cancer cells. Public molecular resources and multi-omics data analysis can prioritize genes of interest for future studies.
Researchers have developed a new method called EvoAug that uses artificial DNA sequences inspired by evolution to train deep neural networks for genome analysis. This approach enables the model to recognize regulatory motifs more accurately, leading to better performance and potential breakthroughs in understanding human health.
A deep learning model has been developed to classify cancer cells into distinct types, enabling accurate prediction of metastatic potential. The tool achieves high accuracy and is simple to use, making it a promising solution for medical practitioners.
More than 3 million sq km of Asian elephant habitat has been lost in 300 years, with suitable habitats cut by nearly two-thirds. The study suggests that the remaining elephant populations may not have adequate habitat areas, setting up a high potential for conflicts with people living in those areas.
A team of researchers from Carnegie Mellon University has developed an AI-based system to help clinicians make decisions quickly and precisely in the ICU. The system, called the AI Clinician Explorer, provides recommendations for treating sepsis based on data from over 18,000 patients.
A mobile application utilizing Python and a single-element ultrasound transducer has been developed for photoacoustic tomography (PAT) image reconstruction. The application successfully reconstructs high-quality images with signal-to-noise ratio values above 30 decibels, making it suitable for point-of-care diagnosis in low-resource se...
A team of experts identified 29 sources of bias in AI/ML models for medical imaging, including data collection, preparation, and deployment. The study provides a comprehensive roadmap for mitigating these biases and ensuring fairness, equity, and trust in AI/ML models.
GPMeta accelerates pathogen detection in metagenomic sequencing (mNGS) tests, achieving higher accuracy while significantly reducing processing time. The approach uses a succinct hash index scheme and multi-GPU support to handle massive data sets.
A study by Drexel University and Vanderbilt University analyzed 82 relevant conversations on Instagram direct messages where teens asked for help, revealing that most disclosures were about mental health concerns. Support was offered in most cases, but specific sets of circumstances led to denial.
A study found that high-quality labeling of images boosts perceptions of training data credibility, leading to increased trust in AI systems. However, biases in the data can reduce trust in certain aspects.
A recent study from Aarhus University found that music used for studying and sleeping share similar characteristics, such as slow tempo and repetitive patterns. The study suggests that these similarities can be attributed to the calming effects of the music on the brain, creating a conducive environment for both tasks.
A machine learning program can spot risky conversations on Instagram by analyzing metadata clues, such as conversation length and participant engagement. The system was 87% accurate in identifying risky chats using sparse and anonymous details from over 17,000 private chats.
A new deep learning-based model estimates breast density with high precision, correlating it to cancer risk. The model's performance is comparable to that of human experts, but it can be trained faster and on smaller datasets.
Max Planck scientists explore the possibilities of artificial intelligence in materials science, discussing how combining physics-based modeling with AI can unlock complex material designs. The research focuses on overcoming limitations of traditional methods and handling sparse, noisy data.
A team of researchers has developed a new human-in-the-loop system to improve the accuracy and interpretability of deep neural networks. The system uses an interactive one-click method for annotating images, reducing the co-occurrence bias inherent in training datasets.
The new dataset provides a 'ranking' of countries contributing most to global warming, with CO2 emissions driving the most warming. Countries like Brazil and Indonesia are rising in their contribution, while industrialised nations see slight declines.
Researchers developed a model to track COVID-19 data, predicting transmission and informing health surveillance systems. The model successfully predicted the spread of COVID-19 in Cali, Colombia, highlighting the importance of high-resolution data in understanding virus dynamics.
Researchers have developed a non-invasive method to track human aging using retinal scans, which are less expensive and more accurate than other aging clocks. The study found that changes in the eye can provide an actionable evaluation of gero-protective therapeutics, offering a new tool for tracking aging.
A national analysis of pediatric hospitalizations from 2009 to 2019 found a significant increase in mental health diagnoses, with attempted suicide being the leading cause. The study highlights the growing importance of addressing mental health concerns in children and adolescents.
Researchers developed an AI tool called SILIC to identify 169 species, including 137 birds, from bird sounds in Yushan National Park. The dataset provides detailed acoustic activity patterns of wildlife across short and long temporal scales.
A machine-learning model was trained on 10 million tweets to infer users' subjective wellbeing. The study found that New Year's Day is the most popular holiday, followed by Defender of the Fatherland Day and International Women's Day. The researchers also discovered gender differences in attitudes towards certain holidays.
Researchers developed AI models based on UNet and MobileNet architectures to analyze standardized abnormalities in CT images, accurately identifying object presence and confidence. These models achieved an absolute percentage error of less than 5 percent, comparable to human professionals.
Researchers found that COVID-19 infections are linked to an increased risk of developing liver problems, acute pancreatitis, and other GI disorders. The study analyzed over 14 million medical records and estimated that SARS-CoV-2 infections have contributed to over 6 million new cases of GI disorders in the US.
A novel AI architecture, relational reasoning network, accurately identifies anatomical landmarks in CT scans for orthodontic treatments. The model learns spatial relationships between landmarks without explicit image segmentation, achieving accuracy comparable to conventional methods.
Researchers seek to develop algorithms providing meaningful explanations for AI decision-making, enabling higher human trust and adoption in fields like science. The project focuses on symbolic reasoning and estimating explanation accuracy, addressing the need for transparent AI systems.
Researchers at MIT developed a technique to improve machine-learning models' reliability without requiring additional data or extensive computing resources. The method uses a simpler companion model to estimate uncertainty, enabling more effective uncertainty quantification.
Researchers tested three common techniques to make algorithms fairer and found that one approach didn't reduce social norm bias at all. They proposed a new technique: a formula to directly measure social norm bias in an algorithm so it can be corrected. This bias can persist even after overt discrimination is removed.
Researchers found that electric car adoption in California was associated with real-world reductions in air pollution and asthma-related emergency room visits. The study also highlighted an 'adoption gap' between low-resource zip codes, pointing to opportunities for environmental justice.
Researchers identified 7 key symptoms of long COVID, including heart issues and joint pain, in a study of 52,461 patients. The findings could help healthcare providers diagnose and treat the condition more effectively.
Dr. Nico Spiller to develop new analysis methods using machine learning to analyze complex brain data related to memory, decision making, and movement. The fellowship aims to provide insights into neurodegenerative diseases such as Alzheimer's and Parkinson's disease.
A new measure called c-value helps researchers choose between techniques based on the chance that a new method is more accurate for a specific dataset. The tool answers questions like whether to use alternative estimation methods despite potential costs and effort.
A recent study published in Nature has discovered several new disease genes and provided new insights into the effects of known genetic factors on disease. The study highlights an underappreciated complexity in dosage effects of genetic variants, challenging traditional Mendelian inheritance laws.