This week’s list of data news highlights covers October 17-23, 2015 and includes articles about a new app that can help people with vision impairments navigate their environment and an artificial intelligent program that can learn a person’s routine through real-life pictures.
The White House has published the final version of its Strategy for American Innovation, which outlines the Obama administration’s efforts to spur technological innovation in the United States. The strategy outlines how the administration plans to administer $146 billion in research and development funding and provide support for innovative technologies related to smart cities, connected vehicles, smart manufacturing, open data, supercomputing, and precision medicine. The strategy also describes the administration’s plans to encourage individuals with data science skills to work in government, rethink regulatory approaches that stifle innovation, and use data to solve tough social challenges.
A San Francisco-based Startup called Amino has launched an online service to help match patients with doctors that best fit their needs based on a database of 3.9 billion healthcare interactions. Amino analyzes data from sources similar to those used by insurance companies and healthcare firms for reimbursement, quality analysis, and fraud detection purposes, such as government agencies, bill collectors, payers, and patient surveys to inform prospective patients about their best possible treatment options. With Amino, users can choose a doctor based on location, insurance carrier, and the doctor’s experience treating a particular condition.
United Arab Emirates Prime Minister Sheikh Mohammed bin Rashid Al Maktoum has issued a law directing government agencies to share non-confidential data with researchers, private sector entities, and each other in an effort to spur Dubai’s transition into a smart city. The law is designed to spur investment in and development of smart city technologies as well as improve the government’s delivery of public services. However, the law does not require government agencies to share their data with the public.
A research partnership between IBM Research and Carnegie Mellon University (CMU) has created a smartphone app called NavCog, which allows people with vision impairments to navigate their environment with their smartphones. NavCog relies on Bluetooth-enabled beacons spread throughout an environment that can indicate precise locations, as well as computer vision algorithms that use images from the smartphone to develop 3D models of the environment, which the app can then use to issue turn-by-turn directions to the user over headphones or by vibrating. CMU plans to install the beacons throughout its campus, and the researchers have made NavCog open source to allow interested developers to experiment with and deploy the technology.
Columbia University and the World Bank have created OpenLandContracts.org to publish data on farmland purchase contracts between governments and investors around the world. The website contains data on 69 land contracts related to palm oil, biofuels, and other agricultural purposes in eight countries, including Ethiopia, Cambodia, and Liberia. The creators of the website say the private nature of these contracts has historically disadvantaged small-scale farmers, and that the lack of disclosure discourages larger firms from incorporating social investments such as sound environmental management practices.
Researchers at the Sanford Burnham Prebys Medical Discovery Institute in San Diego analyzed data from genomic and medical research databases to create a list of genes that contribute to the progression of cancer. The researchers used an algorithm to identify potential relationships between tumor data from 6,000 patients. The data is from the Cancer Genome Atlas, a U.S. government-funded research database, with data on 18,000 protein structures from the Protein Data Bank, an open source research database. The analysis revealed 71 protein structures previously unrecognized as contributing to the progression of cancer, which could help future researchers identify predictive markers for cancer or help develop more targeted cancer drugs.
The White House Office of Management and Budget (OMB) has published a draft of its proposed changes to the A-130 circular, which governs how the U.S. government manages information technology, which has not been updated since 2000. OMB’s proposed changes include the addition of numerous federal directives and executive actions that the White House has issued since the last A-130 modification, including the federal open data policy, established by an OMB memorandum in 2013. The new A-130 would require agencies to continue to make government data accessible to and usable by the public. OMB will accept public comments on the draft via GitHub for 30 days.
Researchers from the Georgia Institute of Technology have developed a computer model capable of identifying activities from real-life images with 83 percent accuracy. The researchers trained a machine learning model called an artificial neural network with 40,000 images taken from wearable cameras, such as Google Glass and GoPro, to represent the wearer’s point of view, tagged with basic contextual information, such as “driving” or “hygiene.” Then, the researchers paired the neural network with an algorithm that can combine data about the time of day each image was taken to help the network identify trends between various activities and make predictions about future events. The researchers expect their model could eventually be useful for personal scheduling applications that incorporate real time data about a user’s activities from wearable devices and cameras.
Costa Rica’s Ministry of Public Works and Transport (MOPT) has announced plans to make data about public transportation publicly available and in the process, become the first public institution in the country to do so. MOPT’s open data portal will initially contain information such as bus routes and fares and taxi fleet composition, and it will gradually include more data over time.
A startup called CareerLabs has launched a service to help jobseekers make more informed decisions about where to apply to for a job by providing users with a wide variety of data about a company that has traditionally been difficult for the average person to access, such as information about corporate financial health, culture, employee turnover, and even political leanings. CareerLabs has compiled 10 million data points on 22 million U.S. companies by analyzing anonymous employee surveys as well as corporate information from government agencies such as the Securities and Exchange Commission.