Bacterial «navigator» accurately tracks where a person has been — scientists have created a new AI tool

Published by Ihor Panchenko

Researchers have created an AI tool that identifies a person’s location based on bacterial samples from their body.

To understand how the new technology works, it is worth recalling how forensic scientists use hair, fibers, gunshot residue, or soil to link a person to a specific location. A new tool from researchers at Lund University in Sweden works in a similar way.

«The human microbiome is constantly changing due to contact with different environments. Tracking microorganisms allows us to understand the spread of disease and identify potential sources of infection,» explains Eran Elhaik, a research biologist at Lund University and the study’s lead author.

Microbial communities have unique geographical features. Some bacteria are globally distributed, while others live only in certain regions or environments. Scientists have focused on bacteria that leave specific «fingerprints» and can point to a specific location.

A large database was used to train the AI: 4135 samples from MetaSUBMetaSUB (The Metagenomics and Metadesign of Subways and Urban Biomes) — is an international database containing genomes of microbes collected from subways and urban environments. collected from subways and various locations in 53 cities, 237 soil samples from 18 countries, and 131 marine samples from nine water bodies. Artificial intelligence was trained to recognize the unique proportions of these «prints» and link them to geographic coordinates. The created system is called Microbiome Geographic Population Structure, or mGPS for short.

The accuracy of the system depends on the number of samples in the database. In well-represented cities, mGPS identified the correct location in 92% of cases. For cities with fewer samples (less than 100), the accuracy was 87%.

To test the accuracy over short distances, the researchers tested the system in the three most studied cities. In Hong Kong, mGPS distinguished between subway stations 172 meters apart. In New York, the technology was able to distinguish between a kiosk and a handrail located at a distance of less than a meter. In London, the accuracy was lower — only half of the samples were correctly linked to geographic clusters. The researchers explain this by the difference in the cleanliness of the metro stations: the samples were collected before the COVID-19 pandemic, when Hong Kong stations were «spotlessly clean» and London stations were — much less well maintained.

«We are just beginning to understand the relationship between microorganisms and the environment,» says Elhaik. The next step is to create microbiome maps of entire cities, which will help in forensic investigations and the study of organisms that live on the streets, in gardens, and on the human body.

The study was published in the journal Genome Biology and Evolution.

Source: New Atlas