Published on October 28th, 2016 | by Joshua New0
10 Bits: the Data News Hotlist
This week’s list of data news highlights covers October 22-28, 2016 and includes articles about the first self-driving truck delivery and a new effort by the U.S. Department of Defense to use AI to develop new treatments for breast cancer.
A team of researchers from the University of London, the University of Sheffield, and the University of Pennsylvania have developed an AI program capable of predicting the outcomes of human rights trials with 79 percent accuracy. The researchers trained their system with historical data about 600 cases from the European Court of Human Rights (ECHR), including case history, relevant legislation, and type of human rights law violation to identify patterns that could suggest a likely outcome. By prioritizing cases based on their most likely outcome, ECHR, which has a large case backlog, could more efficiently judge a larger number of human rights violations.
A group of 10 health-care safety-net clinics, which serve patients regardless of their ability to pay, participating in Minnesota’s Medicaid program have piloted an analytics program that has reduced emergency room use by 18 percent, inpatient hospital use by 8 percent, and spending by 5 percent over 3 years, saving a total of $16.6 million. The program aggregates patient claims data for 30,000 patients and allows the clinics to identify when, where, and how their patients are using their services so staff can make more informed decisions.
Several financial institutions have announced that they are developing artificial intelligence software that can scrutinize financial transactions such as stock trades to catch illegal activity and manipulative behavior. The U.S. Financial Industry Regulatory Authority has announced it is developing machine learning software that can help analyze the 50 billion market events it monitors per day to reduce the high number of false alarms from its traditional analysis software. The Nasdaq exchange is working with a cognitive computing firm called Digital Reasoning that can cross-reference trading data internationally to detect manipulation in near-real time. And the London Stock Exchange has partnered with IBM to use its Watson cognitive computing platform to improve its market surveillance software.
Researchers working for the Oxford University Press used statistical analysis to determine that Shakespeare’s rival, playwright Christopher Marlowe, helped Shakespeare write his three Henry VI plays, settling large amounts of speculation from scholars. The researchers analyzed databases of writings from the Elizabethan period to determine if certain words or combinations of words that appeared in Shakespeare’s plays were common in other author’s’ works, finding that certain types of phrase unique to Marlowe’s works appear in Shakespeare’s writing.
Uber’s autonomous trucking division Otto has made the first-ever delivery of cargo by a self-driving tractor-trailer, delivering 2,000 cases of Budweiser in Colorado Springs, Colorado. The autonomous 18-wheeler truck drove 120 miles on the highway to make the delivery while a human driver sat in the cab, only taking over to drive the truck from the brewery to the highway and off the highway to the delivery site. Anhaeuser-Busch InBevNV, which makes Budweiser, states that it could reduce distribution costs by $50 million per year in the United States by using autonomous trucks.
Messaging software company Slack Technologies has partnered with IBM to allow Slack users to build customized chatbots, called Slackbots, that use Watson’s analytics and natural language processing abilities to better understand and respond to user requests. The companies will create a developer toolkit for its customers that want to build Watson-powered Slackbots, IBM will develop a Slackbot that can help user’s information technology departments respond to user’s questions and resolve problems without having to leave the Slack application, and Slack will upgrade its customer service Slackbot to use Watson.
The U.S. Department of Defense (DoD) has partnered with biopharmaceutical company Berg Health to sequence tissue samples and use Berg’s AI modeling software to generate tissue models that could help lead to new targeted drugs for breast cancer. Berg will sequences 13,600 healthy and diseased tissue samples from a DoD database and its software will attempt to identify certain patterns in the tissue models that could represent biomarkers or drug targets. The initiative is part of the White House’s Cancer Moonshot.
Teva Pharmaceuticals, which makes an Internet-connected smart inhaler, has partnered with IBM Watson to analyze how the weather can influence an Asthma sufferer’s likelihood of an Asthma attack. The companies will develop an app that combines data on a smart inhaler-user’s medication habits as well as local weather data, such as humidity, pollen levels, and temperature, to determine how weather conditions can contribute to risk of an attack, which can be difficult to predict and come on suddenly. Eventually, Teva and IBM hope to be able to send personalized recommendations to Asthma sufferers about elevated risk or if they should ask their doctor to increase their medication.
Trade organizations European Organization of Cosmetic Ingredients Industries and Services (Unities) and the Botanical alliance have partnered to develop a shared database of toxicological profiles and testing data on natural substances, which can help companies better evaluate the safety of their products. The database is designed to help the cosmetics and botanicals industries share knowledge and stay more aware of the potential toxicity of ingredients after the Court of Justice of the European Union recently upheld a ban on animal testing, which helped companies validate the safety of their products.
Researchers at Google Brain, a deep learning project, have developed a system of artificial artificial neural networks capable of developing its their own encryption algorithms without being taught how to do so. The researchers developed the system by programming three neural networks, named Alice, Bob, and Eve, to communicate—Alice sends a secret message to Bob, Bob decodes the message, and Eve attempts to intercept and understand the message. After 15,000 attempts, Alice was able to develop an encryption strategy that Bob could decipher.
Image: Steve Jurvetson.