Hungarian development of MALVINA machine learning method

Researchers from the HUN-REN Szeged Biological Research Centre and the Hungarian Centre for Excellence in Molecular Medicine (HCEMM) have developed a machine learning-based system called MALVINA (Machine Learning-Based Virulence Interaction Analysis) to study bacterial infections at the single-cell level.

Unlike traditional methods that rely on total bacterial quantity, MALVINA uses fluorescent labeling and microscopic image analysis to provide cell-by-cell data. This allows scientists to distinguish specific infection patterns, such as whether bacteria are spreading across many cells or accumulating in a few. The system can simultaneously measure bacterial presence, accumulation levels, and the resulting DNA damage to host cells.

In testing, the method analyzed over 1.3 million human colon epithelial cells infected with various E. coli strains, achieving an 86 percent accuracy rate in distinguishing between infected and uninfected cells. The high-throughput approach is intended to assist in identifying dangerous bacterial strains, testing drug candidates, and developing new probiotic strategies.