This week’s list of data news highlights covers October 10-16, 2015 and includes articles about a Google Street View-style approach to surveying waterways and how Tesla is rolling out a new self-driving Autopilot.
The New York City Metropolitan Transportation Authority (MTA) has begun a test of turn warning and collision avoidance systems for city buses. The 60-day pilot will rely on bus-mounted sensors installed on six buses that constantly monitor the speed of the bus and can detect potential collisions to the front or sides of a bus. When there is a high risk of an accident, the system will warn the driver to make necessary adjustments. If the pilot is successful, MTA will expand the program to 100 buses for further testing next year.
Tesla Motors has released an autopilot mode for its Model S cars which allows the vehicles to use data from their many numerous cameras and sensors to make driving decision with little to no input from human drivers. The autopilot system is not fully autonomous, but it can automatically detect parking spaces and parallel park on its own, slow to a stop if it detects the driver has fallen asleep, and better warn the driver of potential safety issues such as if the car is drifting out of its lane.
A nonprofit called the Freshwater Trust has partnered with Google to more efficiently survey the waterways in the United States to support conservation efforts. Freshwater Trust is using Google Trekker, a raft-mounted version of technology used by Google Street View cars, to take photographs of waterways. In addition, it is using sensors to measure oxygen and phosphorous levels, as well as water temperature. The organization analyzes the data it collects along with climate data, land rights information, and soil tests, to recommend where the government should focus its conservation efforts.
Apple’s ResearchKit, a platform that allows Apple device owners to contribute their data to clinical research, is taking on three new projects to study epilepsy, autism, and melanoma. Researchers from Johns Hopkins School of medicine created EpiWatch to collect biometric and qualitative survey data from epilepsy patients using the Apple Watch with the goal of developing a method to detect seizures before they happen. Researchers at Duke University have developed the Autism & Beyond app to use ResearchKit to map a child’s facial expression during a reaction to certain stimuli, which could lead to better autism diagnoses. And researchers from Oregon Health and Science University have launched the Mole Mapper app to allow users to take pictures of moles at regular intervals to study changes over time, which could lead to a melanoma diagnosis, as well as contribute their pictures to a research database.
The National Comprehensive Cancer Network (NCCN), a nonprofit consortium of cancer hospitals, has issued guidelines to provide cancer patients with better data about the cost of drugs used in their treatment. The guidelines, which are available to all hospitals, will rank drugs using the same scale used to help patients compare drug toxicity and effectiveness. By providing a common ranking system, the guidelines will help doctors and their patients make better informed treatment decisions.
The U.S. National Institute of Standards and Technology (NIST) has published a guide to solve common usability challenges associated with electronic health records (EHR) that jeopardize patient safety. NIST identified three major problems: poor identification, consistency, or integrity of data; lost data; and data stored in disparate locations. The agency tailored its guidance to address these challenges using empirical data gathered from analysis of multiple EHR systems, on-site observations, EHR user surveys, and expert reviews of EHRs.
Facebook has begun to use artificial intelligence algorithms to help interpret photos and videos into spoken words to make the website more accessible for blind users. Screen reading technology, which many blind people use to navigate websites, cannot interpret photos and videos, but now users can click on photos or videos to have Facebook’s algorithms describe their contents and context.
The U.S. Federal Register and the Government Publishing Office have partnered to digitize all of the Federal Register’s 1.8 million pages of government regulations. Currently, the public can only access Federal Register documents online if they were published after 1994. All records dating back to 1936 will be available in digital format, and each document will have searchable metadata related to the regulation it covers. The Federal Register expects to complete the project by next year.
A researcher at the University of Nevada, Reno has developed a prototype system to help blind runners exercise by using drones to help them navigate a track. The system uses a drone equipped with two cameras—one that can identify and follow lines on a track and another that focuses on the runner—that emits a noise to help guide runners along the track. The drone can also detect when a runner changes speed and adjust its own speed accordingly.
Researchers at the Georgia Institute of Technology have developed an algorithm that allows unbalanced robots to identify the safest way to fall down to avoid damaging themselves. The algorithm relies on motion sensors and cameras on a robot to calculate how the robot should move to create contact points with the ground that disperse the momentum of the fall. Eventually, the researchers hope to eventually devise methods to help robots avoid hurting people when they fall.
Image: Office of Naval Research.