A new research from IZS-Teramo aims at refining and harmonizing the most advanced methods of bacterial pathogen DNA analysis through the application of Artificial Intelligence using Machine Learning techniques. The goal is to develop swift tools that can effectively respond to foodborne infectious diseases.


Like all living forms, bacteria are constantly adapting to their environment, in order to increase their chances of survival and reproduction. This evolutionary process leads to the development of specific characteristics that enable them to effectively respond to diverse environmental conditions. So their genetic heritage contains critical information that can reveal significant aspects of the life and history of that microorganism, including its natural habitat.


Genomic information is crucial when it comes to the study of bacterial pathogens, such as Listeria monocytogenes, responsible for listeriosis, a disease that can pose a severe threat to individuals with weakened immune system, such as the elderly, pregnant women, or those suffering from chronic and degenerative diseases. With thousands of deaths worldwide, listeriosis is considered one of the most serious animal-origin zoonoses.


Hence, when listeriosis cases surface, there is the need to identify, as quickly as possible, the contaminated food source that caused the infection, tracing back the production chain in order to intervene promptly and limit the number of cases. This process, known in technical terms as "source attribution", was the focus of a new scientific study by the Bioinformatics Unit of IZS-Teramo. The study compared different Machine Learning parameters that can predict the food origin of the bacterium starting from its genome. These findings were published in the prestigious scientific journal BMC Genomics.


“In the specific case of listeriosis – says Nicolas Radomski, the last author of the scientific work - we must keep in mind that the disease can have a long incubation time. So it becomes difficult to identify the food responsible for the infection by simply questioning the patient. Most often, he does not remember what he ate a week before. Furthermore, he would typically have consumed a variety of foods. Using advanced genomic sequencing and Machine Learning techniques, researchers can now assign probabilities regarding the food origin of the pathogen and therefore the foods at risk”.


Traditional epidemiological investigations following cases of listeriosis can take a long time, in some cases even months. This consists of tracking source of infection through checking and analysis of all foods commonly consumed by patients. The methodologies highlighted in this scientific work, instead, provide crucial leads that can expedite the investigations, allowing for a rapid focus on particular foods for further analysis.


“These techniques – the researcher concludes –possess a wider and more flexible application potential. In particular, they can be pivotal in managing and understanding other critical and urgent phenotypes, such as the growing problem of antimicrobial resistance. By implementing and optimizing such approaches, Artificial Intelligence can significantly accelerate and improve the discovery of resistance-related mutations, thus contributing to the development of more effective strategies to combat the phenomenon.”




Pierluigi Castelli, Andrea De Ruvo, Andrea Bucciacchio, Nicola D'Alterio, Cesare Camma, Adriano Di Pasquale and Nicolas Radomski (2023) Harmonization of supervised machine learning practices for efficient source attribution of Listeria monocytogenes based on genomic data. 2023, BMC Genomics, 24(560):1-19


Nicolas Radomski


© IZSAM November 2023

Istituto Zooprofilattico Sperimentale

dell'Abruzzo e del Molise "G. Caporale"


Campo Boario | 64100 TERAMO | ITALIA

Telefono 0039.0861.3321 | Fax 0039.0861.332251


Posta elettronica certificata:

Partita IVA: 00060330677

Codice Fiscale: 80006470670