During Ignite 2020, Microsoft stopped its efforts to detect pathogens before they spread. Building on the experience of five years of the Premonition epidemiological project led by Microsoft Healthcare, the company expects to use a combination of robotics, cloud infrastructure and sensors to monitor and select carriers and identify known and emerging biological threats.
In the coming weeks Intuition the team will be available in advance by joining a larger group of partners, including the National Science Foundation Convergence Accelerator Program, Johns Hopkins University, Vanderbilt University, the University of Pittsburgh, and the Institute of Health Metrics and Assessment at the University of Washington. In addition, Microsoft says it will expand its partnership with Bayer to investigate vector-borne diseases and the role of autonomous sensor networks in determining biological treatments.
Experts agree that the prevention of health crises such as a coronavirus pandemic is an early detection of potentially harmful pathogens. Pathogen detection gives researchers, health organizations, and governments time to develop new treatments, prepare public responses, and reduce exposure.
However, pathogens move through the environment in difficult-to-monitor ways. The U.S. Centers for Disease Control and Prevention has estimated that 60-75% of new infectious diseases – such as zika, dengue fever and West Nile – are caused by pathogens that go from animals to humans. Humans are usually exposed through the excrement or slaughter of host animals, but bites are another vector for transmission.
Five years ago, Microsoft’s Central Engineering Team, a hardware lab that provides electromechanical engineering expertise to Microsoft research teams, helped found Premonition. From the outset, the aim of the project was to expand surveillance to detect disease threats at an early stage.
The central engineering team developed a foot-height 64-compartment robot that monitored for insects capable of transmitting pathogens and collected samples from other animals. In fact, most terrestrial animals are arthropods (such as insects, spiders, and mites) that pollinate crops, provide food for larger animals, and process biodegradable in the environment. However, arthropods, especially mosquitoes, are one of the major vectors of human disease. For example, the mosquito species Aedes aegypti carries Zika and dengue fever, while Culex quinquefasciatus spreads West Nile virus.
The robot, which has survived three prototypes, adaptively lures, identifies and selectively captures mosquito species where it is deployed. Using carbon dioxide and spring traps, battery-powered microprocessors and camera-mounted mosquito chambers, the robot launches algorithms to identify species of artificial intelligence to capture “medically important” mosquitoes with more than 90% accuracy. The algorithms look at the patterns emitted by the mosquito-wielding wing, illuminating the cameras with infrared light. Microsoft engineers collected wing images and, along with Johns Hopkins researchers, identified varieties and taught algorithms throughout the pilot deployment.
Work on a “next-generation library” of classifiers has begun, a Microsoft spokesman told VentureBeat. Hariss County Public Health in Texas transported mosquito species to describe Microsoft campuses. The Premonition team anticipates that the classifiers will be installed at new facilities in Harris County by 2021. Winter. Microsoft expects to report more next summer and fall as Harris County public health entomologists evaluate the accuracy of the classifiers.
The systems inside and outside the cameras digitize the behavior of mosquitoes in high definition, describing how they move around. The cameras, designed to withstand rain, wind and other elements, also tell investigators how long each mosquito has been trapped and what the temperature, wind, light level and humidity were when the mosquito entered.
The data recorded by the robot is transmitted to Azure for analysis. In the cloud, the Premonition metagenomics pipeline evaluates organisms and viruses in a sequential environmental sample. Algorithms and a large database of reference genomes allow the system to study sequencing pathogens and detect associated viruses and bacteria, while looking for links between pathogens and their potential hosts.
Microsoft says Premonition can scan a sample using trillions of genetic signatures to distinguish not only viruses but also bacteria, fungi and higher levels of life. This also makes it possible to identify the types of animals fed by mosquitoes.
Initially tested with Harris County Public Health in Houston, Texas, during the Zika handover in 2016, Microsoft says Premonition has since been tested in South Florida and Tanzania.
“We used to assign shorter pilots to lead engineering and science for several months. We always removed these traps from the environment when the experiments were completed because they were not intended for long-term deployment, ”the spokesman said. “However, our newly announced partnership with Harris County Public Health will showcase hundreds of devices in selected locations to allow continuous monitoring of the environment over a long period of time (e.g., years). We will work with Harris County Public Health to further disseminate knowledge on how to make optimal use of these data flows and select locations to improve public health operations. ”
In addition to replicating and installing robotic traps, the Premonition team is working on ways to use drones to monitor the spread of disease and mitigate disease outbreaks. In the short term, scientists hope to use unmanned aerial vehicles with computer vision technology to look for areas where mosquitoes that can carry diseases tend to choose.
The Premonition team is also developing a “next generation” device that is optimized for public health programs and will be announced in 2020. At the end or 2021. In the beginning. Microsoft then plans to describe its early access program in detail and allow a choice between public and private health and service services. organizations to deploy the entire Premonition stack, including software desktops and cloud computing credits, traps, and anthropoid digitization tools.
Talk director Ethan Jackson finally hopes thousands of traps and drones will be deployed around the world. Teamwork would complement initiatives such as the Global Virome Project and the US National Institutes of Health recently released a pathogen tracking program that aims to find and sort out every virus that can pose a threat to humanity.
Other efforts to identify potentially dangerous pathogens are continuing. Truly health-oriented subsidiary of Google’s parent company, Alphabet, closed the veil Coordination project The project uses AI algorithms to quickly distinguish Aedes aegypti mosquitoes from males and females. This year, Verily managed to eradicate disease-causing mosquitoes from three test sites in California’s Central Valley. 2017 Entomologists Stanford University has developed an artificial intelligence-based sound-aware smartphone app that can differentiate between mosquito species based on the noise made by their wings. And Puerto Rico Fabrics put together a set of images and labels to teach a computer vision algorithm that recognizes mosquitoes.
“There are more than 3,600 species of mosquitoes, but if we can catch the right ones and perform a metagenesis sequence in all the DNA and RNA in them, we can see diseases along the way,” Jackson said. New York time.