Pharmacokinetic forecasting on PK-DB and R pharmacokinetic datasets (leave-one-individual-out)
0.174Caffeine ErrorFlowPK
Evaluation results
| Method | Links | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| FlowPKProtocol=leave-one-individual-out, Conditioning=past-conditioned (first four PK samples) + study context (S) + dosing (u)2026.04 | 0.174 | — | 0.963 | 0.375 | 0.245 | 0.404 | 0.295 | 0.593 | 0.309 | 0.681 | 0.495 | 0.184 | 0.421 | 0.422 | 0.293 | 1.026 | 0.585 | 0.07 | 1.156 | 0.265 | 0.273 | |
| NLMEProtocol=leave-one-individual-out, Conditioning=past-conditioned (first four PK samples) + study context (S) + dosing (u)2026.04 | 0.294 | — | 0.752 | 0.3 | 0.582 | 0.343 | 0.596 | 1.3 | 0.299 | 0.479 | 0.383 | 0.732 | 0.723 | — | — | — | — | — | — | — | — | |
| AICMETProtocol=leave-one-individual-out, Conditioning=past-conditioned (first four PK samples) + study context (S) + dosing (u)2026.04 | 0.399 | — | 0.595 | 0.691 | 0.536 | 0.475 | 0.264 | 0.936 | 0.352 | 0.339 | 0.558 | 0.274 | 0.517 | 0.491 | 0.335 | 1.188 | 0.563 | 0.076 | 1.082 | 0.307 | 0.312 | |
| NODEProtocol=leave-one-individual-out, Conditioning=past-conditioned (first four PK samples) + study context (S) + dosing (u)2026.04 | 0.432 | — | 0.605 | 0.661 | 0.35 | 0.554 | 0.352 | 1.003 | 0.582 | 0.693 | 0.535 | 0.327 | 0.578 | 0.737 | 0.353 | 1.474 | 0.834 | 0.049 | 1.39 | 0.297 | 0.287 | |
| TransformerProtocol=leave-one-individual-out, Conditioning=past-conditioned (first four PK samples) + study context (S) + dosing (u)2026.04 | 0.451 | — | 0.395 | 0.683 | 0.429 | 0.513 | 0.288 | 0.842 | 0.447 | 0.709 | 0.502 | 0.356 | 0.678 | 0.567 | 0.354 | 1.286 | 0.733 | 0.069 | 1.054 | 0.357 | 0.413 |