This week’s list of data news highlights covers July 9-15, 2016 and includes articles about San Francisco’s new smart highway project and a machine learning system that could help detect Alzheimer’s disease.
The Obama Administration has launched the Advanced Wireless Research Initiative to accelerate the research and deployment of advanced wireless infrastructure to support the Internet of Things and 5G technologies. The National Science Foundation (NSF) and private sector groups will invest $85 million to develop four city-scale wireless testing platforms to allow researchers to conduct large-scale tests of advanced communications technologies. NSF will also invest $250 million over the next seven years on advanced proof-of-concept wireless projects to test on these city platforms. Also as part of the initiative, several other agencies including the Defense Advanced Research Projects Agency and the National Institute of Standards and Technologies, as well as private sector and academic groups, will carry out research projects focused on connectivity issues such as spectrum sharing, sensor networks, and device interference.
The California Department of Transportation (Caltrans) has begun initial work on its SMART Corridor project, which will integrate connected technologies into a 20 mile span of Interstate 80 to improve traffic flows and roadway safety. Caltrans will install Internet-connected traffic lights and signage with monitoring technologies that can dynamically adjust traffic patterns and speed limits based on traffic conditions, as well as sensors in on-ramps that can provide traffic monitoring stations will real-time traffic data. Caltrans expects the SMART Corridor to be substantially more cost effective at relieving the congestion and safety problems that plague Interstate 80 than widening the road, which could take decades and cost hundreds of millions of dollars.
A Cambridge, Massachusetts startup called Sense has developed a consumer device that can monitor a household’s electricity usage in real time and identify specific appliances that are draining too much energy. Once installed inside an electrical service panel, the device uses current sensors to capture data about energy usage and transmits this data to a cloud-based analytics platform to disaggregate this data and identify the specific energy-use patterns of different appliances. The device can accurately disaggregate 80 percent of a home’s energy use and consumers can use this data to reduce energy use or replace power-hungry appliances. Sense will also use this data to develop a system that can provide personalized energy-savings recommendations.
Researchers at Vrije University Medical Center in Amsterdam have developed machine learning algorithms that can analyze magnetic resonance imaging (MRI) brain scans and detect signs of Alzheimer’s disease based on physical cues. The researchers trained the algorithms to identify differences in perfusion, the process of brain tissue absorbing blood, in MRIs, as decreased perfusion reduces the amount of oxygen and nutrients in the brain and is linked with Alzheimer’s disease and other forms of dementia. In tests, the algorithms could differentiate between a patient suffering from Alzheimer’s and patients suffering from other forms of dementia with between 82 and 90 percent accuracy. This approach could eventually help doctors identify early warning signs of dementia likely to progress to Alzheimer’s disease and prompt earlier treatment.
The American Heart Association (AHA) has partnered with Amazon Web Services (AWS) to store and analyze patients’ genetic, environmental, and lifestyle data in the cloud to advance research into precision cardiovascular treatments and reduce costs. By combining different sources of patient data, researchers can gain new insights into the factors that contribute to cardiovascular disease and develop more effective, personalized treatments. AHA has launched a portfolio of grants to focus on improving analytical techniques for cardiovascular medicine and AWS will provide grant recipients with access to AHA data and analytics services.
Berkeley-based startup Civil Maps has developed machine learning software that can analyze light detection and radar (LIDAR) sensor data from self-driving cars to create accurate maps at a fraction of their regular size. Self-driving cars use LIDAR to sense the world around them, but only by generating huge amounts of data—less than half a square mile of LIDAR-mapped environment can take up several gigabytes of storage space, which poses challenges for self-driving cars to store, share, and update large amounts of maps. Civil Maps uses machine learning software to strip away unnecessary information from LIDAR maps, leaving only the data essential for a car’s navigation, to greatly reduce their size without reducing their functionality. Civil Maps’ software can convert LIDAR maps of 300 lane miles into just eight megabytes.
The Human Cancer Models Initiative, a project supported by research institutes in the United States, UK, and the Netherlands, has launched a pilot project to compile 1,000 3D models of cancer, called cell lines, to help cancer researchers around the world better understand the huge amount of variation in cancer types. The 1,000 new models, which will double the amount of cancer models accessible to researchers, will be matched with clinical data about donor patients as well as data about how patients responded to different treatments. With these models, researchers can more easily model the progression of different types of cancer and develop new treatments, and doctors can more quickly identify treatments that might benefit their patients.
Smart home device company Nest has added the Nest Cam Outdoor security camera, which uses artificial intelligence software to identify potential security threats, to its lineup of connected home devices. Unlike most security cameras that rely on simple motion detection, Nest Cam Outdoor uses a machine learning system to differentiate between non-threatening movements and sounds, such as a moving tree branch, and humans, to notify owners of a potential security threat. Nest Cam Outdoor also links with other connected home products to promote home security, such as automatically turning on indoor lighting to mimic someone being home, if it detects a person outside late at night.
Pål Sundsøy, a researcher at Norwegian-based telecommunications provider Telenor, has developed a method to determine literacy rates in developing countries by using machine learning algorithms to analyze mobile phone call records. Sundsøy matched survey data on thousands of mobile phone users in a developing country with Telenor’s call records to build a model network of each user to determine who they call, where they call from, and other mobile usage data. Then, using machine learning algorithms, Sundsøy was able to identify phone usage patterns that serve as reliable indicators of literacy, such as where users spend their time, differences in volume between incoming and outgoing text messages, and size of social network, and predict whether or not an individual is literate with 70 percent accuracy. This technique could help provide international aid efforts with granular data about areas with low literacy rates.
A new online social network called Data.World has launched to help data scientists connect with each other and access and analyze datasets. The developers of Data.World created the platform to make it easier for data scientists to find useful data, which they believe has hindered valuable research, and provide them with a freely accessible and useful infrastructure for data sharing. Data.World is working with the U.S. National Science Foundation to host the Census Bureau’s open data on the platform and allow users to combine public data sources with private datasets for their own analysis, collaborate with other users working on the same datasets, and host and share their own datasets.