Method Profile
Baseline data from an isolated human lung generate an untreated forecast for comparison with its measured response to alteplase.
Evaluate short-term organ-specific drug effects with a computational paired reference, alongside conventional controls.
Hybrid physiological and machine-learning forecasts trained on human EVLP data, coupled to real perfused donor lungs.
A first-hour baseline predicts second- and third-hour untreated values; compare the physical lung after 20 mg alteplase with that forecast and with a historical untreated cohort.
Development and test evidence
951 documented EVLP cases from a 1,000-case review; modality-specific complete-case samples; 45 separate clinical test cases from the same center; alteplase-treated cases excluded from training.
Drug experiment
16 alteplase-treated lungs, 2 excluded for insufficient perfusion follow-up; 6 suitable and 8 unsuitable lungs analyzed.
Keep clinical test, simulated illustrative and drug-treated cohorts separate. The paired untreated trajectory is predicted, so the drug contrast depends on forecast calibration.
Independent same-center forecast test and real-world treatment-use comparison, with suitable and unsuitable subgroups analyzed separately.
45 clinical test lungs; 50 simulated illustrative profiles; 14 analyzed treated lungs (6 suitable, 8 unsuitable) after 2 of 16 exclusions.
Each treated lung’s untreated digital forecast; randomly selected historical untreated group of 6 lungs.
Suitable six: conventional comparisons were nonsignificant, while paired PAP at 2 hours was lower (P = 0.031). Edema was not significantly different (P = 0.063). Unsuitable eight did not show the same pressure-lowering response.
An ex vivo lung model forecasts untreated function from baseline data. In six alteplase-treated lungs suitable for transplantation, paired comparison found lower pulmonary arterial pressure after 2 hours; eight unsuitable lungs did not show the same benefit.
