Method Profile
Molecular structure is used to estimate milk-transfer parameters for a lactation PBPK model, with predictions compared against observed clinical milk-to-plasma ratios.
Research prioritization of medicine exposure during breastfeeding using milk concentrations, relative infant dose and pharmacological context.
Support vector regression estimates milk partitioning and permeability–surface-area parameters, which feed a postpartum whole-body PBPK model.
Compare clinical milk-to-plasma ratio predictions with a conventional equation-based lactation PBPK benchmark, then screen drug exposure scenarios.
Inputs
Molecular fingerprints, physicochemical descriptors, milk-to-plasma partitioning and permeability–surface-area parameters.
Evaluation
Observed clinical milk-to-plasma ratios and a conventional equation-based PBPK benchmark.
Compare the proportion of predictions within two-, three- and fivefold error; assess the effects of maternal clearance, blood-to-plasma ratio and milk partitioning.
Clinical milk-to-plasma ratios are compared with predictions from RISKmilk and a conventional lactation PBPK model.
Drugs with observed clinical milk-to-plasma ratios, plus marketed-drug screening scenarios.
Prediction-error distributions for RISKmilk versus the conventional model.
Twofold agreement increased from 29.6% to 66.4%. Four of 14 drugs with predicted milk-to-plasma ratios of at least 1 exceeded 10% relative infant dose in the maximum scenario.
A structure-based framework improved milk-to-plasma ratio predictions against a conventional lactation model. Its screening results separate predicted milk enrichment from relative infant dose and pharmacological risk.
