Marcelo Sousa, a biochemist at the University of Colorado Boulder, had spent ten years trying to crack a particularly tricky puzzle. Sousa and his team had collected reams of experimental data on a single bacterial protein linked to antibiotic resistance. Working out its structure, they hoped, would help to find inhibitors that could stop that resistance from building. But, year after year, the puzzle remained unsolved. Within 15 minutes, DeepMind's machine learning system had solved the structure.
New York, July 14: A centralised repository of COVID-19 health records built by US researchers, last year, has been helpful in tracing the progression of the disease over time and could eventually be used as the basis for decision-making tools. The National COVID-19 Cohort Collaborative (N3C) is a centralised, harmonised, high-granularity electronic health record repository that is the largest, most representative COVID-19 cohort to date. 'Treatment With Blood Thinners May Reduce Death in COVID-19 Patients', Says Study This multicenter data set can support robust evidence-based development of predictive and diagnostic tools and inform clinical care and policy, said a team of researchers from those including at Universities of Colorado, Michigan, Rochester Medical Center, and Johns Hopkins. The cohort study, published in the JAMA Network, used data from 34 medical centers and included over 1 million adults -- 174,568 who tested positive for COVID-19 and 1,133,848 who tested negative between January 2020 and December 2020. "This cohort study found that COVID-19 mortality decreased over time during 2020 and that patient demographic characteristics and comorbidities were associated with higher clinical severity," said Tellen D. Bennett, from Department of Pediatrics at Colorado's School of Medicine.
Machine learning is gaining popularity across scientific and technical fields, but it's often not clear to researchers, especially young scientists, how they can apply these methods in their work. In many ways, ESS present ideal use cases for ML applications because the problems being addressed--like climate change, weather forecasting, and natural hazards assessment--are globally important; the data are often freely available, voluminous, and of high quality; and computational resources required to develop ML models are steadily becoming more affordable. Free computational languages and ML code libraries are also now available (e.g., scikit-learn, PyTorch, and TensorFlow), contributing to making entry barriers lower than ever. Nevertheless, our experience has been that many young scientists and students interested in applying ML techniques to ESS data do not have a clear sense of how to do so. An ML algorithm can be thought of broadly as a mathematical function containing many free parameters (thousands or even millions) that takes inputs (features) and maps those features into one or more outputs (targets).
We have encountered units and elementary dimensional analysis in our high school science classes. For instance, the mass of an object is expressed in kilograms (kg). Likewise, length is expressed using meters (m) and time in seconds (s). Other physical quantities such as acceleration has dimensions m s-2 (derived from its definition), whereas force has dimensions kg m s-2. The latter arises from Newton's second law that states that force (F) is equal to the mass (m) times the acceleration (a).
Neural networks are increasingly used in environmental science applications. Furthermore, neural network models are trained by minimizing a loss function, and it is crucial to choose the loss function very carefully for environmental science applications, as it determines what exactly is being optimized. Standard loss functions do not cover all the needs of the environmental sciences, which makes it important for scientists to be able to develop their own custom loss functions so that they can implement many of the classic performance measures already developed in environmental science, including measures developed for spatial model verification. However, there are very few resources available that cover the basics of custom loss function development comprehensively, and to the best of our knowledge none that focus on the needs of environmental scientists. This document seeks to fill this gap by providing a guide on how to write custom loss functions targeted toward environmental science applications. Topics include the basics of writing custom loss functions, common pitfalls, functions to use in loss functions, examples such as fractions skill score as loss function, how to incorporate physical constraints, discrete and soft discretization, and concepts such as focal, robust, and adaptive loss. While examples are currently provided in this guide for Python with Keras and the TensorFlow backend, the basic concepts also apply to other environments, such as Python with PyTorch. Similarly, while the sample loss functions provided here are from meteorology, these are just examples of how to create custom loss functions. Other fields in the environmental sciences have very similar needs for custom loss functions, e.g., for evaluating spatial forecasts effectively, and the concepts discussed here can be applied there as well. All code samples are provided in a GitHub repository.
A team of researchers at the Max Planck Institute for Intelligent Systems in Germany and at the University of Boulder in Colorado in the US has now found a new way to exploit the principles of spiders' joints to drive articulated robots without any bulky components and connectors, which weigh down the robot and reduce portability and speed. Their slender and lightweight simple structures impress by enabling a robot to jump 10 times its height. At the end of May, the team's work titled "Spider-inspired electrohydraulic actuators for fast, soft-actuated joints" was published in Advanced Science. The high performance is enabled by Spider-inspired Electrohydraulic Soft-actuated joints -- SES joints in short. The joints can be used in many different configurations -- not just when creating an arachno-bot.
Jack Morrison and Isaac Roberts (far left and right) previously cofounded and sold 3D scanning company Replica Labs to Occipital. There they met electrical engineer Davis Foster (center), with whom they went on to cofound Scythe Robotics. Self-driving cars get all the hype. But while the category continues to face a long and uncertain path to commercialization, a burgeoning crop of autonomous vehicles is already hitting the market. The latest is Scythe Robotics, a Boulder, Colorado-based company that announced today it is launching a zero-emission, autonomous lawn mower backed by $18.6 million from Inspired Capital, True Ventures and more.
LITERATURE UPDATE May 20, 2021 - May 26, 2021 Literature search terms: biomech* & locomot* Publications are classified by BiomchBERT, a neural network trained on past Biomch-L Literature Updates. BiomchBERT is managed by Ryan Alcantara, a PhD Candidate at the University of Colorado Boulder. Each publication has a score (out of 100%) reflecting how confident BiomchBERT is that the publication belongs in a particular category (top 2 shown). Risteski P, Jagrić M, Pavin N, Tolić IM, Current biology: CB. (76.3% CELLULAR/SUBCELLULAR; 4.7% MUSCLE) Physical analysis reveals distinct responses of human bronchial epithelial cells to guanidine and isothiazolinone biocides. Kwon TY, Jeong J, Park E, Cho Y, Lim D, Ko UH, Shin JH, Choi J, Toxicology and applied pharmacology.
When the constitution of the American Association for the Advancement of Science was revised in 1946, its statement of objectives contained new language: “…to increase public understanding and appreciation of the importance and promise of the methods of science in human progress.” The association has since fulfilled that charge in diverse sectors, including policy, education, and public engagement, to make science more relatable and relevant to the public. Making science relatable also requires a variety of engagement strategies, including facilitating in-depth discussions with local policy leaders, translating technical language into digestible summaries for the classroom, and promoting science role models. In the case of the AAAS Center for Scientific Evidence in Public Issues or EPI Center, for instance, a successful part of bringing clear and actionable scientific advice to policy-makers has been encouraging discussions among a broad group of experts and policy peers. During meetings organized by the EPI Center this year, city council members, mayors, water engineers, and local utility managers joined scientists to discuss perand polyfluoroalkyl substances or PFAS, synthetic chemicals found in drinking water systems. At least two PFAS have been associated with increased rates of some cancers and thyroid disease. The EPI Center provides nontechnical syntheses of topics for policy-makers, “but one thing we have seen is that examples from their peers that have implemented and used the scientific evidence are much more valuable and easier to understand,” said Kathryn McGrath, communications director for the center. Whether the focus is clean water or voting technology or hydraulic fracturing, the EPI Center strives to make the science of these topics relatable by talking with the public and policy-makers to find out exactly what information would be helpful for them. The discussions allow city council members, for instance, “to ask the science experts what they need to know to go back to their communities and regions and take action on some of these issues,” McGrath said. AAAS's Local Science Engagement Network, a grassroots platform that nurtures local and state science advocates for climate and energy policy, has also found success with local partnerships. In Colorado, Missouri, and Georgia, LSENs work with organizations in each state that “have a good sense of policy landscapes as well as the cultural and scientific landscapes in those areas,” said Daniel Barry, local and state advocacy director and head of LSEN at AAAS. LSENs offer an avenue for engagement and advocacy that AAAS members have been asking for, by connecting scientists with their own elected representatives on the local, state, and federal levels. As both constituents and neutral, honest brokers of scientific information, LSEN participants can be a key resource when legislatures grapple with the more local implications of climate change, such as modernizing the state power grid, said Barry. “They can step up and say, ‘Science, that's what I do, and I live here in this community. I know how to get you the science you need.’” LSEN members also condense technical research into locally relevant analyses in plain English for business leaders and citizens. So far in 2021, Missouri LSEN partner MOST Policy Initiative has produced more than 80 such “science notes” about pending state legislation. Among AAAS's numerous education efforts to make science more relevant is Science in the Classroom, an initiative that annotates and provides additional resources to accompany research papers from the Science family of journals. The goal is to make scientific papers more accessible to high school, community college, and undergraduate students, while putting a face on the papers' authors in communities with little exposure to working scientists, said program director Suzanne Thurston. The popular resource had more than 1 million page views in the past 3 years, and the hunger for accessible scientific content during a pandemic year led to a 50% increase in total site visits in 2020 compared to 2019. The program also offers professional development workshops to educators, researchers, and annotators. By showcasing a range of authors and annotators, Science in the Classroom helps “to expose students to diversity within STEM and demonstrates what ‘actual living scientists’ look like,” said Thurston, who serves as a program director in AAAS's Inclusive STEM Ecosystems for Equity and Diversity (ISEED). The IF/THEN Ambassador program, led by AAAS's Center for Public Engagement with Science and Technology, was another recent effort to show off the diverse faces of science, by highlighting 125 women in STEM as role models for middle school girls. Lyda Hill Philanthropies, which funds the IF/THEN initiative, wanted to work with AAAS on the ambassador program after the association's success with other public engagement initiatives such as the AAAS Mass Media Science & Engineering Fellowship and the Leshner Leadership Institute for Public Engagement with Science, said Emily Therese Cloyd, director of the AAAS Center for Public Engagement with Science and Technology. The ambassador program was distinguished by its emphasis on increasing visibility for women in STEM who demonstrate how science is involved in everyday careers beyond the traditional lab, said Cloyd. “We're moving beyond scientists who work at an academic institution and thinking about the ways that a video game designer or a fashion designer might be using STEM every day.” AAAS is committed to making science relatable and relevant for everyone from policy-makers to educators to students. It is at the core of the organization's mission and will continue to be a top priority for years to come.
The Aspen Fire Protection District is piloting new technology that will keep an eye -- er, AI, rather -- on wildfires this summer using artificial intelligence technology and strategically placed cameras, the district announced Saturday. The system uses specialized cameras at specific vantage points to monitor the skyline, coupled with artificial intelligence and intuitive software technology from wildfire tech company Pano AI to detect, locate and communicate wildfire threats almost instantly, according to a news release. "Pano's platform uses mountaintop cameras, artificial intelligence, and intuitive software to automatically detect the first wisps of smoke and put real-time fire images in the hands of first responders and emergency personnel, all with the goal of detecting flare-ups earlier and enabling a faster response before they become large infernos," the release states. Cameras stationed on Pitkin County communications towers will continuously rotate to capture 360-degree views of the area; Pano AI software will process that imagery in real time to detect smoke and alert dispatchers or appropriate agencies. When multiple cameras capture the same smoke wisps, the software can use triangulation to pinpoint the location, "helping response crews coordinate a faster, more targeted response," according to the release.