Microbial traits have minimal power for predicting death from bloodstream infection
Gilliland, R.; Valdes-Tresanco, M. E.; Plakhotnyk, A.; Westlund, A.; Smith, J. T.; Pham, T. M.; Mortimer, T. D.; Walker, B. J.; Earl, A. M.; Grad, Y. H.; MacDonald, M. E.; Lewis, I. A.; PRIME Consortium,
Show abstract
Bloodstream infection (BSI) is a life-threatening condition where the risk of dying is affected by both patient- and pathogen-related factors. Certain microbial traits (e.g., virulence genes) are thought to affect survival rates from BSIs, but the magnitude of their impact on human mortality remains unclear. To address this, we analyzed the microbial genomes, quantitative proteomes, and electronic medical records (EMRs) from 35,746 BSIs from Calgary, Canada (2006-2022). We show that EMR-based machine learning models accurately predict 30-day mortality (AUC=0.83), but surprisingly, integrating >70,000 microbial genes/proteins does not improve this performance. We demonstrate that microbial factors co-segregate with specific patient cohorts and are therefore not independent predictors of risk. We conclude that microbial virulence factors have minimal impact on mortality rates from BSIs.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Dynamics of gut mucosal colonisation with extended spectrum beta-lactamase producing Enterobacterales in Malawi 92%
- Antimicrobial exposure in sexual networks drives divergent evolution in modern gonococci 92%
- The major pathogen Haemophilus influenzae experiences pervasive recombination and purifying selection at local and global scales 90%
Similar papers in this journal
- A Genome-based Model to Predict the Virulence of Pseudomonas aeruginosa Isolates 93%
- Aggregation of nontuberculous mycobacteria is regulated by carbon:nitrogen balance 92%
- Streptococcus mitis bacteriocins drive contact-dependent lysis of S. pneumoniae facilitating transformation in multispecies environments 91%
Similar papers in this journal
Similar papers in this journal
- Integrating theory and machine learning to reveal determinants of plasmid copy number 92%
- Prophages and satellite prophages are widespread among Streptococcus species and may play a role in pneumococcal pathogenesis 92%
- Species-level microbial sequence classification is improved by source-environment information 92%
Similar papers in this journal
- Peering Inside the Black Box: Explainable AI to Interpret Advanced Computer Vision Fungal Pathogen Prediction 91%
- Identifying likely transmission pairs with pathogen sequence data using Kolmogorov Forward Equations; an application to M.bovis in cattle and badgers 91%
- Pseudolysogeny-mediated evolutionary trade-offs favor phage therapy by limiting antibiotic resistance and virulence in Cutibacterium acnes 90%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.