10 Bits: the Data News Hotlist
This week’s list of data news highlights covers October 14-20, 2017, and includes articles about an AI system that can prevent unnecessary surgery for breast cancer and a chatbot that uses AI to help people manage depression.
Self-driving car company Waymo has started to work with local law enforcement agencies and emergency responders to develop protocols for when a self-driving vehicle gets into a collision. When one of Waymo’s cars crashes, it will automatically report details about the crash to Waymo and local law enforcement agencies, and Waymo is also conducting on-site training to help first responders to recognize and access its vehicles. Waymo is also working with the Chandler, Arizona police department to train its cars to recognize and respond to emergency sirens by stopping or pulling over, and in the event of a crash, Waymo’s cars can use audio sensors to detect where emergency sirens are coming from so it can respond appropriately.
Researchers at the Massachusetts Institute of Technology, Massachusetts General Hospital, and Harvard Medical School have developed an AI system that determine if lesions identified in a biopsy after a mammogram will eventually develop into cancer with 97 percent accuracy. Diagnosing breast cancer typically involve analyzing lesions identified during mammograms that appear to have a high risk of becoming malignant tumors, however 90 percent of these tumors turn out to be benign and these false positives can result in a high number of unnecessary surgeries. By helping doctors better understand the likelihood a high-risk lesion will become cancerous, the system could reduce the number of unnecessary surgeries by over 30 percent when compared to existing diagnostic techniques.
A startup called Woebot has developed a chatbot with the same name that uses AI to converse with users and help them manage symptoms of depression. Woebot uses natural language processing to interpret conversations, track users’ moods, and provide information about cognitive behavioral therapy users can apply in their daily lives. In a study, Woebot was able to reduce depression symptoms in students in two weeks of use.
The U.S. National Institutes of Health (NIH) has launched a five-year public-private partnership with 11 pharmaceutical companies called the Partnership for Accelerating Cancer Therapies (PACT) to accelerate the developments of precision medicine treatments as part of its Cancer Moonshot initiative. PACT, with $160 million in funding from NIH and $55 million from participating companies, will oversee research into identifying biomarkers that can signal how a patient will respond to different immunotherapy treatments. PACT’s goal is to reduce the time it takes to find a successful immunotherapy treatment for a patient by half.
The U.S. Department of Homeland Security (DHS) has launched a competition called the Hidden Signals Challenge to incentivize the development of monitoring systems that use open data to detect biothreats, such as anthrax or smallpox outbreaks. DHS will award $300,000 in prizes to participants that develop systems that can pull in open data, such as air sampling data from DHS or the Centers for Disease Control’s Early Aberration Reporting System, which compiles data about health risks from health departments and emergency calls, to detect biothreats within 10 days of first exposure.
The Prime Minister of the United Arab Emirates has launched the UAE Strategy for Artificial Intelligence to accelerate the development of AI and spur public sector adoption of the technology. The strategy focuses on five themes: establishing an AI workforce team in government; carrying out programs to conduct workshops and increase the government’s familiarity with AI; developing AI training programs for government officials; integrating AI into public services, including healthcare and security, to increase productivity and efficiency; and and creating a strategy to promote the safe use of AI.
Startup Built Robotics has developed an autonomous track loader, a type of tractor used for excavation in construction, with a suite of sensors designed to withstand the hazards of construction sites. The track loader uses specialized LIDAR sensors that can operate in environments with high amounts of vibration and withstand impacts. Another sensor also allows the vehicle to measure the amount of material it digs up. And to navigate, Built Robotics developed a system that combines traditional GPS navigation with on-site location detection technology so the track loader can navigate and dig with centimeter-level accuracy.
A team of physicists led by researchers at the California Institute of Technology in Pasadena have successfully used a quantum computer to sort through the huge amounts of data generated by the Large Hadron Collider (LHC) to identify signs of the Higgs boson particle. Physicists have used the LHC to prove the existence of the higgs boson in 2012, but the focus of the new research was to prove that it is possible to perform the same kind of analysis using machine learning techniques on a quantum computer. Though conventional approaches are currently more effective, the research suggests that quantum computing could surpass existing techniques as the technology progresses.
The British Broadcasting Corporation (BBC) has launched a five-year research partnership with eight universities in the United Kingdom called the Data Science Research Partnership to study machine learning in the media industry. The Data Science Research Partnership will focus on four categories of potential AI applications, including understanding audiences, understanding content, curation and personalization, and “content of the future,” which includes predicting what kinds of content will be popular in the future and new forms of data journalism.
Researchers at the University of Washington have launched an open database called MyGene2 to serve as a platform for patients, researchers, and clinicians to share genetic data about rare Mendelian disease—diseases caused by a single inherited gene, such as sickle-cell anemia and cystic fibrosis. Analyzing genetic data can help researchers develop new insights and potentially new treatments for these diseases, but because these diseases are rare, compiling enough useful information can be challenging. MyGene2 allows patients to upload their genetic data along with other relevant medical data, such as their symptoms, as well as access relevant information about their disease.