A new technology from Israel aims to use AI to take the online casino gamble out of figuring out how to predict the spread of coronavirus. Researchers have developed a method that employs Artificial Intelligence to help identify and monitor coronavirus outbreaks and possibly to predict future danger zones. The method uses algorithms to analyze the data so authorities can predict where COVID-19 is most likely to develop and monitor that region.
AI (Artificial Intelligence) is a field of computer science that focuses on creating intelligent machines that work and react like humans. In short, it aims to create intelligent computers which are meant to facilitate:
- Speech recognition
A computer that is programmed for artificial intelligence is programmed for:
- Ability to manipulate and move objects
If a computer has enough information relating to the world it can be programmed to act and react like a human. The programmer is responsible for initiating common sense, reasoning and problem-solving power in machines. There are different components of AI including robotics, machine perception, deep learning and machine learning.
AI and Coronavirus Pandemic
Israeli researchers are working on a method to help identify, monitor and predict future outbreak zones where the coronavirus may next occur. The project is being jointly developed by the Prof. Yuval Dor of the Hebrew University of Jerusalem and professors Eran Segal and Benjamin Giger of the Weizmann Institute of Science of Rehovot. The two institutions are working in conjunction with Israel’s Health Ministry on the initiative.
The system uses questionnaires which are filled out by the general public. The data is then analyzed by algorithms and artificial intelligence. The goal is to be able to predict where the virus is most likely to spread. Then, authorities can focus on that area to monitor the development of virus symptoms more closely.
A pilot was launched in Israel during the week of March 15th. About 60,000 people filled out the questionnaires.
After a preliminary analysis of the data, scientists and researchers were able to detect a significant increase in symptoms in locations that they were able to identify as those that had been visited by confirmed corona patients.
The program’s developers say that, by using their AI program, authorities will be able to predict which locales are most likely to suffer an outbreak. This will allow them to centre their efforts on the locations in which they are most needed.
The Health Ministry is using this data to show the numbers of home quarantine cases in each Israeli community. The data gives them a better understanding of how and where the coronavirus spreads. The Israeli Health Ministry has publicized a map that shows cities with 5000 or more residents. Each location is marked with a circle that reflects the number of people who are quarantined in that city or town. Even though the map doesn’t provide the number of confirmed cases in each area, through the map people can see how the virus spreads. Some people are in quarantine after they return from abroad but others are also in isolation after they had contact with a confirmed coronavirus patient.
The map put out by the Health Ministry also reflects government statistics on those locales where the largest number of quarantined people live. They include those areas that are under a blanket quarantine aimed to halt the spread of the virus. Some of the largest contaminated areas in Israel are Efrat where there are 20 confirmed coronavirus cases
Topping the list is the West Bank settlement of Efrat, which has a population of 9,200 residents and 20 confirmed cases of coronavirus. 9.5% off all the settlement’s residents are in-home quarantine. Another community hard-hit by the coronavirus is Kiryat Ya’arim which has a population of approximately 6000 residents and at least 25 confirmed coronavirus cases, most of whom were infected during a community event. 1,500 of the town’s resident have entered home quarantine. Communities with the lowest number of people in home quarantine are in Arab areas where, on average, there’s less than one person in quarantine for every 1,000 residents.
Relying on AI
The era of AI and big data is revolutionizing tracking and forecasting the path of infectious disease outbreaks including that of the coronavirus. The earlier and more detailed the information, the better health authorities can determine where to screen for infected people and how to allocate resources. Much of the information from which the system feeds come from the world’s largest Internet companies, including Google. Google supplies search keyword and location data to some pandemic-detection programs.
Facebook is also involved, sharing posts mentioning the coronavirus from Facebook Groups and Instagram and aggregated data about users’ movements. Anonymized data from Twitter and other social media sites also fuels the algorithms which typically run on servers managed by Google, Microsoft and Amazon using chips specifically designed for AI (as opposed to the monitoring firms’ own computers).
Pumping huge amounts of information into machine-learning systems is no guarantee of success. In 2013, Google tried to forecast the severity of seasonal flu outbreaks but wildly overestimated the 2013 cycle after searchers’ search for health care information fooled the system into forecasting a higher rate of infection than actually occurred.
Companies developing pandemic-detection systems must ensure that they don’t get misled by hysteria and focus only on relevant bits of information. That’s why the systems rely on humans to examine each case and adjust the sources of information upon which their technology relies.
It’s also important to use fresh data. In the early days of the pandemic, initial simulations of how the coronavirus could spread relied on past air travel itineraries. When movements within China and to and from China changed, due to government restrictions, travel patterns changed. In that case, Metabiota, a San Francisco startup using AI to detect pandemics updated its archive of historical passenger information with real-time location data from travellers’ mobile phones.
It’s also important, in such a case, to use actual medical data instead of news reports and online chatter. The Kinsa smart thermometers have an app that helps users decide when to see a doctor. Those thermometers provide clues about the spread of disease. Kinsa claims that it exceeds the accuracy of the CDC flu forecast. They hope to develop a system that will predict flu outbreaks up to three months in advance.