Table 1.
Selected pAST methods targeting organisms relevant to PJI.
| Organism Group | Organism | Input Data Type | Decision Algorithm Type | Data Input to Decision Algorithm |
Decision Algorithm |
Validation Strategy | Validation Dataset Sizes | Accuracy on Validation Datasets [# Compounds] |
Year | Ref. |
|---|---|---|---|---|---|---|---|---|---|---|
| Staphylococci | S. aureus | Gene | Rule | AMR markers in reads or assemblies | ResFinder DB lookup | Multiple independent validation datasets | ~80 | 96% and 97% [8] | 2020 | [28] |
| Gene | Rule | AMR markers in reads or assemblies | Custom AMR DB lookup | Independent validation dataset | 470 | (Sensitivity/specificity of 99.1%/99.6%) [12] | 2015 | [29] | ||
| Sequence | ML | DNA k-mer counts from assemblies | AdaBoost | 10× cross-validation | 11 | 99.5% [1] | 2016 | [43] | ||
| Sequence | ML | DNA k-mer counts from reads or assemblies | Set Covering Machine | 80%/20% data split | ~330 | 95%–99% [10] | 2019 | [51] | ||
| Gram-negatives | K. pneumoniae | Sequence | ML | DNA k-mer counts from assemblies | XGBoost | 10× cross-validation | ~160 | (VME/ME of 4%/7%) [21] | 2018 | [35] |
| E. coli | Gene | Rule | AMR markers in reads or assemblies | ResFinder DB lookup | Multiple independent validation datasets | 95, 99, 390 | 97%, 98% and 95% [16] | 2020 | [28] | |
| Gene | ML | gene content, isolation year, population structure | Multiple | 80%/20% data split | 387 | 91% [11] | 2018 | [60] | ||
| Sequence | ML | DNA k-mer counts from reads or assemblies | Set Covering Machine | 80%/20% data split | ~300 | 80%–98% [16] | 2019 | [51] | ||
| P. aeruginosa | Gene | ML | AMR markers, gene content, gene expression data | Support Vector Machine | 80%/20% data split | ~80 | (Sensitivity of 81%–91%) [4] | 2020 | [61] | |
| Sequence | ML | DNA k-mer counts from reads or assemblies | Set Covering Machine | 80%/20% data split | ~100 | 73% and 95% [4] | 2019 | [51] | ||
| Sequence | ML | DNA k-mer presence/absence patterns | Regression | 75%/25% data split | 48 | 88% [1] | 2018 | [63] | ||
| Enterococci | E. faecium | Sequence | ML | DNA k-mer counts from reads or assemblies | Set Covering Machine | 80%/20% data split | 27 | 100% [1] | 2019 | [51] |
| Gene | Rule | AMR markers in reads or assemblies | ResFinder DB lookup | Multiple independent validation datasets | 50 and 56 | 93% and 96% [8] | 2020 | [28] | ||
| E. faecalis | Gene | Rule | AMR markers in reads or assemblies | ResFinder DB lookup | Independent validation dataset | 50 | 97% [5] | 2020 | [28] |