Nice People Networking

Dominican invents AI device to predict disease outbreaks

26-year old Dominican Rainier Mallol has invented a medical epidemiological device with artificial intelligence (AIME) that predicts where the next outbreaks of dengue, zika and chikungunya are most likely to occur. The invention turned Mallol into one of 35 winners of Innovators under 35 in Latin America 2017 of the MIT Technology Review in Spanish.

MIT Technology Review explains that medical epidemiology studies the patterns of onset and spread of infectious diseases, but always at posteriori. Mallol used available data from the study of the conditions that favor the outbreaks to identify the zones that have a greater risk due to their characteristics. His platform enabled the detailed analysis of large volumes of data by an artificial intelligence (AI) algorithm, a tool that is more powerful and faster than human experts can calculate. The AI is shown to accurately predict the likelihood of an outbreak occurring at a particular location three months in advance.

“The project currently focuses on three mosquito-borne diseases: dengue, Zika and Chikungunya,” says Mallol. In the case of dengue, the effectiveness achieved exceeds 88%. To do this calculation, Mallol fed its platform with historical data from which it made the predictions. Then he had to compare his results only with the outbreaks that actually occurred after three months. Having this level of precision with that margin in advance would allow authorities to make more effective use of resources, such as more effectively targeting fumigation campaigns.

After conducting pilot studies in Brazil and the Philippines, AIME has reached an agreement with the government of the Malaysian state of Penang to use its platform. “Each case reported by physicians in health centers is digitized and reported immediately to the system, a subsystem collects data related to the case from other sources (NASA, Google, weather, etc.), up to a total of 246 variables,” explains Mallol. With these data, including the domicile of the affected and the date of the first symptom, the AI model updates its prediction of the percentage of risk of generating an outbreak (two or more cases in an area of 400 meters in 14 days) for each location.

In addition to the prediction map, the platform generates a control panel that displays, in real time, various strategic analytics, such as the total history of cases, age ranges of the affected, ethnic background and symptoms presented, to name a few examples. “In 2016 four scientists took almost a year to map all the cases of 2014, grouped by months, with AI that work was done in three hours,” asserts the young Dominican, as reported by the MIT Technology Review.

Source: DR1, Technologyreview

Oct 11, 2017

Category: DR News |

Leave a comment

You must be logged in to post a comment.

Last updated December 17, 2017 at 1:23 AM
stats for wordpress
View Statistics Report
Facebook Twitter