A model can perform well by a stated criterion while retaining clinically relevant uncertainty.
What the publication shows
The modeling study developed a network including midazolam and several CYP3A or transporter interaction partners. It used published studies for development and separate interaction predictions. More than ninety percent of predicted exposure and peak-concentration ratios were within twofold of observed ratios.
Limits of the evidence
Twofold agreement is a model-evaluation criterion, not exact patient prediction. Published-study selection, parameter assumptions and unmodeled physiology affect transferability. The model did not directly measure sedation or determine an individual’s concentration from a known dose.
Application to expert review
For a pharmacokinetic opinion, report the validation population, prediction interval and relevant route or interaction. Do not equate a successful average exposure-ratio prediction with certainty about clinical impairment. Transparent model documentation allows assumptions to be examined rather than hidden behind a numerical output.
Source and related reading
Source: Publication PMC6202474. The case-review applications above are interpretive analysis.
Related reading: Older-Adult Psychotropic Review Found Agreement on Long-Term Use Alongside Barriers to Change.
Leave a Reply