Machine learning successfully simulates and predicts the spread of influenza

In a new study on machine learning published in the British journal Nature Communications, a team of U.S. scientists reported that machine learning-based analysis of anonymous cell phone data can successfully simulate and predict the spread of the viral disease, influenza. Current research suggests that this mobile map can accurately predict the spread of influenza in New York City and Australia, and that there may or may be potential to monitor new coronary pneumonia in the future.The spread of viral diseases in populations depends on the interaction between infected and unseeded people. Current model data used to predict the spread of disease in a city or country present sparse and inaccurate problems
, such as commuter surveys or online search data.To get a denser data set, Google researcher Adam Sadilok and colleagues collected anonymous tracking data from Android phones that turned on Location History and used machine learning to split it into individual "trips" to create a crowd movement map. Using a model of infectious disease transmission calibrated based on hospital registration and inspection data, they used the mobile map to successfully simulate "forecasting" flu activity in and around New York City from 2016 to 2017.The team found that the model performed better than the commonly used standard forecasting model and was similar to using commuter survey data, but that it was known to be more expensive to collect. They also simulated "forecasting" flu transmission in Australia during the 2016 flu season. Although Australia's population is sparser and flu dynamics are different, the model can still predict the peaks and troughs of influenza with greater detail.Existing high-resolution mobile data comes from cell phone call logs that are provider-specific and generally do not reflect cross-border or cross-border movements. Location data have no such limitations and therefore have more potential for monitoring the spread of disease over long distances. Currently, these data are lacking in integrity because mobile data for children and the elderly with low smartphone usage is not included. Despite these limitations, the team demonstrated the potential of using mobile phone data to predict the spread of epidemics. (Journalist Zhang Mengran)Editor-in-Chief circle pointPeople often difficult to predict when the virus will enter the human body, lurk, quietly spread in the crowd, and then outbreak of a war. Predicting the epidemic of infectious diseases is a very necessary but difficult task in densely populated metropolises. Research suggests that mobile phone data, coupled with artificial intelligence, may have predictive potential. But technology is never a thing of the last. Infectious diseases are predicted and powerful interventions are needed to "kill" them in their infancy. Controlling the source of infection, cutting off transmission routes, and protecting susceptible populations are three ancient but effective methods. But to do this, not only rely on artificial intelligence, but also rely on human intelligence and decision- (Science and Technology Daily)