Home BlogWeekly News 10 Bits: the Data News Hotlist

10 Bits: the Data News Hotlist

by Cassidy Chansirik
Image of a cow.

This week’s list of data news highlights covers October 3, 2020 – October 9,  2020 and includes articles about predicting heart failure using machine learning and building the UK’s fastest supercomputer for AI research in healthcare. 

1. Predicting Heart Failure Using Machine Learning

Researchers at the Massachusetts Institute of Technology are using machine learning to identify patients at risk of heart failure by detecting excess fluid build-up in their lungs, an early warning sign of heart failure. To identify the severity of the condition, the team trained its algorithm to interpret over 300,000 lung x-rays and radiologist reports. When tested, the algorithm accurately diagnosed 90 percent of patients with severe fluid build-up. 

2. Predicting Cows at Risk of Disease Using Machine Learning 

Researchers from Washington & Lee University in Virginia and pharmaceutical company Elanco Animal Health, have developed a set of predictive models to identify cattle at risk of bovine respiratory disease (BRD). BRD is a contagious infection that kills thousands of cattle in the United States every year, costing the meat production industry approximately $1 billion. To predict which cattle are at-risk, the researchers’ model uses data that is already electronically collected through ear tags, such as temperature, and maps where the cattle have been and which cattle have been in contact with one another. 

3. Blocking Offensive Comments Using AI 

Instagram is using AI to automatically block offensive comments on new posts. The platform already automatically deletes comments that violate its community guidelines, but now it will use AI to block comments users have previously flagged as offensive from appearing on their posts. Currently, Instagram is piloting these features for comments written in English, Spanish, French, Portuguese, Chinese, Russian, and Arabic. 

4. Building the UK’s Fastest Supercomputer for AI Research in Healthcare 

Nvidia, a company that develops graphic chips, is building the UK’s fastest supercomputer for AI research in healthcare and drug discovery. The supercomputer will serve as a hub of innovation, supporting research in large-scale healthcare and data-science problems for AI startups, medical institutions, and universities. When complete, the system is forecast to be among the top 30 fastest and top 3 most energy-efficient supercomputers in the world. 

5. Discovering Craters on Mars with AI 

Scientists and AI researchers at NASA’s Jet Propulsion Laboratory have developed an AI tool that can identify craters on Mars caused by meteor impacts, which could contain geological clues about life on the planet. To identify these craters, the team trained the AI tool to scan over 100,000 low-resolution images. While it takes a researcher an average of 40 minutes to scan a single image, the AI was able to analyze images in 5 seconds, which can help researchers investigate more meteor impacts for clues about life on the planet. 

6. Predicting the Susceptibility of Animals Contracting COVID-19

Scientists at Vanderbilt University in Tennessee have developed an algorithm that predicts how susceptible animals are to contracting COVID-19. The researchers first identified that there are five amino acid sites on the ACE2 protein, which the coronavirus uses to gain bodily entry, that affects whether an animal is able to contract the disease. The scientists then developed an algorithm that analyzes an animal’s entire ACE2 sequence for these sites and predicts how at-risk they are. The researchers found that horses, camels, and the Chinese horseshoe bat are at a high-risk of contracting the virus. 

7. Identifying Cancerous Cells Using Machine Learning

Researchers from the University of California, Irvine have developed biochips, miniature laboratory materials used in biochemical analysis, that use machine learning to differentiate between cancerous and healthy cells. To do this, the system monitors differences in the electrical properties of these cells using electrodes and then uses machine learning to process and analyze the large amount of data the system produces. The system can identify diseased cells 96 percent of the time.

8. Improving Quantum Chemistry Using Machine Learning 

Researchers from the California Institute of Technology have developed a machine learning tool that helps predict molecular properties of compounds to aid quantum chemistry research, the study of chemical properties and processes at the quantum scale. Unlike current methods, the tool uses neural networks to mathematically map atoms and molecules in a way that is naturally aligned to quantum mechanics. This allows quantum calculations to be more easily performed, enabling researchers to better predict properties such as the structure of molecules, the way in which they will react, whether they are soluble in water, and how they will bind to a protein.

9. Increasing Harvesting Yield Using AI

Deere & Company, a manufacturer of agricultural machinery, has created harvesting combines with high-resolution cameras and sensors linked to AI software to help farmers improve their yield and reduce costs. The system monitors grain as farmers collect it and adjusts settings on the combine in real-time to maximize how much grain they chop from each stalk and to minimize waste. The system can harvest a field 45 percent faster than the company’s older equipment that did not have the automated system, and it uses 20 percent less fuel.

10. Screening Stool for Indicators of Cardiovascular Disease 

Researchers from the University of Toledo in Ohio have developed an algorithm that can screen individuals for cardiovascular disease (CVD) by analyzing the bacterial makeup present in their stool sample. First, the team found that higher levels of certain bacteria were present in the stools of individuals with CVD. Then they trained a machine learning model to screen individuals for CVD based on whether this bacterial signature was present in their stool samples. Those individuals that the system flags as at-risk are referred to their doctor for additional testing and therapeutic intervention. 

Image: Ryan McGuire

You may also like

Show Buttons
Hide Buttons