Man
Professional
- Messages
- 3,088
- Reaction score
- 631
- Points
- 113
A new way to track a person's path with amazing accuracy.
Swedish researchers have developed an AI capable of determining a person's route through the bacteria they have collected during their movements.
Every day we unwittingly collect a lot of microorganisms from the environment. Unlike human DNA, which remains unchanged, our microbiome is constantly changing when it comes into contact with different surfaces and environments.
The system, called mGPS (Microbiome Geographic Population Structure), works in a similar way to traditional forensics techniques, where hair, fibers, traces of gunpowder, and soil samples are used to link a person to a specific place, object, or event.
The head of the study, Eran Elhaik, explains that bacterial navigation will help not only solve crimes, but also track the spread of diseases, find sources of infections and localize the emergence of antibiotic-resistant microorganisms.
To create the technology, the researchers processed huge data sets. Microbiome samples from the urban environment, soil, and marine ecosystems were used. Based on them, the neural network was trained to recognize the unique proportions of microbial fingerprints and associate them with geographical coordinates.
To be more precise, a dataset of 4135 MetaSUB samples was analyzed - a collection of microbial genomes collected in the subway and urban spaces of 53 cities. In addition, 237 soil samples from 18 countries and 131 marine samples from nine bodies of water.
The first tests impressed the researchers: the algorithm accurately identified the cities from where 92% of the samples came. To check whether the amount of data affects the result, the scientists conducted a separate test in 31 cities where fewer than 100 samples were collected. The accuracy practically did not decrease, reaching 87%.
The team then focused on the three megacities where they had accumulated the most information: New York, Hong Kong and London. And here the results did not disappoint: in Hong Kong, the technology distinguished samples from two subway stations at a distance of only 172 meters from each other.
In New York, the accuracy was even higher - the tool distinguished the microbiomes of the kiosk and the handrail at a distance of less than a meter. With London, it turned out to be a little more complicated: the location was named correctly for only half of the samples. Scientists explain this by the condition of the local metro. Samples were collected before the COVID-19 pandemic, and the stations of the British capital turned out to be significantly dirtier than the "impeccably clean" stations of Hong Kong.
According to the creators, the accuracy of the system will increase as new microbiome data is added. Now the team intends to expand the study: if previously samples were collected only in the subway and on the streets, now they plan to create detailed bacterial profiles for all types of urban spaces - from parks to shopping centers.
Source
Swedish researchers have developed an AI capable of determining a person's route through the bacteria they have collected during their movements.
Every day we unwittingly collect a lot of microorganisms from the environment. Unlike human DNA, which remains unchanged, our microbiome is constantly changing when it comes into contact with different surfaces and environments.
The system, called mGPS (Microbiome Geographic Population Structure), works in a similar way to traditional forensics techniques, where hair, fibers, traces of gunpowder, and soil samples are used to link a person to a specific place, object, or event.
The head of the study, Eran Elhaik, explains that bacterial navigation will help not only solve crimes, but also track the spread of diseases, find sources of infections and localize the emergence of antibiotic-resistant microorganisms.
To create the technology, the researchers processed huge data sets. Microbiome samples from the urban environment, soil, and marine ecosystems were used. Based on them, the neural network was trained to recognize the unique proportions of microbial fingerprints and associate them with geographical coordinates.
To be more precise, a dataset of 4135 MetaSUB samples was analyzed - a collection of microbial genomes collected in the subway and urban spaces of 53 cities. In addition, 237 soil samples from 18 countries and 131 marine samples from nine bodies of water.
The first tests impressed the researchers: the algorithm accurately identified the cities from where 92% of the samples came. To check whether the amount of data affects the result, the scientists conducted a separate test in 31 cities where fewer than 100 samples were collected. The accuracy practically did not decrease, reaching 87%.
The team then focused on the three megacities where they had accumulated the most information: New York, Hong Kong and London. And here the results did not disappoint: in Hong Kong, the technology distinguished samples from two subway stations at a distance of only 172 meters from each other.
In New York, the accuracy was even higher - the tool distinguished the microbiomes of the kiosk and the handrail at a distance of less than a meter. With London, it turned out to be a little more complicated: the location was named correctly for only half of the samples. Scientists explain this by the condition of the local metro. Samples were collected before the COVID-19 pandemic, and the stations of the British capital turned out to be significantly dirtier than the "impeccably clean" stations of Hong Kong.
According to the creators, the accuracy of the system will increase as new microbiome data is added. Now the team intends to expand the study: if previously samples were collected only in the subway and on the streets, now they plan to create detailed bacterial profiles for all types of urban spaces - from parks to shopping centers.
Source
