The abstract is read at the publisher; the summary is JClub's.
Random Forest source attribution using the resistome of nearly 1200 Acinetobacter baumannii genomes identifies humans and companion animals well (F1 weighted 0.744), but soil and white stork predictions fall below 50%.
- Why it matters: Source attribution is important for tracking transmission across human and non-human reservoirs, yet critical pathogens like Acinetobacter baumannii have lacked resistome-based machine learning approaches.
- What they did: They compared 168 parameter combinations, selecting Random Forest with 5-fold cross-validation and chi2 feature selection, then assessed class-dependent performance across four sources (humans, companion animals, soil, white storks).
- The result: Wrong predictions are linked to multi-host lineages, and key ARGs include tetgenes among top contributors for companion animals, supporting resistome-based models for predicting source associations.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.