Catalyst Twin module
Melatonin,
one oral
dose.
Plasma melatonin after one immediate-release oral dose, in a one-compartment model whose parameters are values papers print, but for an absorption rate fitted to one paper’s time to peak, tested against published summaries from studies it was not built from.
- Engine
- melatonin.twin, version 2
- Built from
- 4 papers
- Held out
- 7 studies 13 of 18 inside
- Validation
- holds
Front panel
computed at buildCh 01Outcome a range for a reference population, never a point
PeakCmax
2,460 to 8,400 pg/mL, median 4,610 pg/mL
Time of peakafter the dose
34 to 60 min, median 45 min
Half-lifeelimination
30 to 121 min, median 56 min
Total exposureAUC to infinity
4,320 to 28,800 pg·h/mL, median 11,300 pg·h/mL
Ch 02Parameters every one with its source
| Block | Parameter | Value | Read from | Source |
|---|---|---|---|---|
| Gut | Absorption half-life (kₐ = ln 2 / t), fitted so the typical subject peaks at 41 min | 19.1 min | “Tmax of 10 mg orally administrated melatonin was 41 min in a study” | PMID 35413810fitted to this record |
| Gut | Absorption half-life as printed; not used, as it peaks below every held-out time to peak | 6 min | “Melatonin administered orally has a half-life of absorption of 6 min” | PMID 35413810study record, not used |
| Gut | Oral bioavailability, F (typical subject) | 15% | “its oral bioavailability—referred to as a 2–4 mg dosage—is approximately 15%” | PMID 38791203study record |
| Gut | F between people, drawn log-uniform | 9%–33% | “The half-life of melatonin by both oral and intravenous routes is approximately 45 min (28–126 min), and the oral bioavailability is low with significant individual variability (9–33%)” | PMID 35409137study record |
| Plasma | Volume of distribution | 42 L | “V p (L) | 42 (35; 48)” | PMID 36046952study record |
| Elimination | Elimination half-life (typical subject) | 45 min | “The half-life of melatonin by both oral and intravenous routes is approximately 45 min (28–126 min), and the oral bioavailability is low with significant individual variability (9–33%)” | PMID 35409137study record |
| Elimination | Half-life between people, drawn log-uniform | 28–126 min | “The half-life of melatonin by both oral and intravenous routes is approximately 45 min (28–126 min), and the oral bioavailability is low with significant individual variability (9–33%)” | PMID 35409137study record |
Transcribed from 4 papers in the graph’s study records, but for one fitted value: the absorption half-life, 19.1 min, solved so the typical subject peaks at 41 min, the time to peak PMID 35413810 prints after 10 mg orally. The 6 min absorption half-life the same paper prints peaks at 20 min, below every held-out time to peak, so it is recorded and not used. Between people, bioavailability and the half-life are drawn log-uniform across the ranges their paper prints; 300 virtual subjects, seed 1.
No dose is advised. The ranges are for a stated reference population: a simulation from published values, not a prediction for any person.
Validation held out
Each column is one published summary from a study the model was not built from. The green line is the 5th to 95th percentile of 1,000 virtual subjects; the dot is what the paper reported, orange where it falls outside that range. Both are divided by the median prediction, so ×1 is an exact match and the dashed lines mark twofold either way.
- Studies
- 7
- Inside the range
- 13 of 18 (72%)
- Cmax
- 2 of 4
- tmax
- 4 of 7
- AUC0–∞
- 2 of 2
- t½
- 5 of 5
- Within twofold
- 16 of 18
- Half-life doubled
- 6 of 18 (33%)
The control: the same model with every half-life doubled keeps 6 of 18 inside, against 13. The fit: the absorption half-life is fitted to one time to peak, from PMID 35413810, a paper that built the model and is outside the test. None of the 18 held-out summaries was fitted to, so the share is the same counted either way: 13 of 18 with every held-out summary, and 13 of 18 with the fitted-to ones taken out, since there are none. The held-out times to peak test whether that one fitted value carries to other studies; version 1, with the printed 6 min, kept none of them inside. The rule, fixed before the first run: the validation holds when at least half of all summaries, and at least half of each kind, fall inside. It holds.
Not modelled
A measured absorption rate. The absorption half-life is fitted to one published time to peak, 41 min after 10 mg orally (PMID 35413810), so the typical subject’s time to peak is set by the fit, not predicted; the 6 min that paper prints as the absorption half-life peaks at 20 min, below every held-out time to peak. Whether the study it cites also lies behind a held-out review cannot be told from the record.
A delay before absorption. The model absorbs from the moment of the dose; the lag time PMID 36046952 estimates carries no unit in the record, so it is not used.
Between-person differences in absorption and in volume. The papers give a range only for bioavailability and for the half-life, so only those two vary.
First-pass metabolism as a process. Bioavailability is one fraction of the dose, the same at every dose.
Metabolism to 6-hydroxymelatonin and its sulphate, and anything that changes it: CYP1A2 inhibitors such as fluvoxamine, smoking, liver disease.
The body’s own melatonin. The curve is the dose alone, added to nothing; night-time secretion is not drawn.
Food. Every arm is fasted or does not say; one review reports time to peak moving from 0.75 to 3 hours with food (PMID 36660543).
Prolonged-release, sublingual and transdermal forms. One immediate-release oral dose only, never repeated.
Age, sex, body size and disease. The volume is one number in litres, not scaled to a body.
Any effect: sleep, the MT1 and MT2 receptors, the circadian clock. The model stops at plasma.
Left out of the test
| Source | Records | Why |
|---|---|---|
| PMID 35413810 | absorption half-life, tmax, half-life after intravenous dosing, bioavailability | Built the model. |
| PMID 38791203 | bioavailability | Built the model. |
| PMID 35409137 | half-life and bioavailability, with their ranges | Built the model. |
| PMID 36046952 | volume; Cmax and tmax at 20 and 50 mg, in sepsis and in healthy subjects | Built the model; the arms are also patients with sepsis. |
| PMID 22348451 | Cmax, tmax, half-life, CL/F, V/F at 0.4 and 4 mg | Older adults with insomnia, a group named as such. |
| PMID 18423009 | Cmax, tmax, AUC, half-life, CL/F at 10 mg | Critically ill patients. |
| PMID 28368352 | AUC at 3 mg against placebo | Critically ill patients. |
| PMID 19606092 | Cmax and tmax from a 2.1 mg patch | Transdermal, not oral. |
| PMID 37438493 | Cmax and tmax of a sublingual spray and a prolonged-release tablet | Sublingual and prolonged-release, not an oral immediate-release dose. |
| PMID 35176989 | tmax and half-life of slow-release formulations | Slow-release. |
| PMID 36660543 | tmax with food; bioavailability | Fed arm; bioavailability is not a compared summary. |
| PMID 36976674 | bioavailability | Not a compared summary. |
| PMID 31033013 | tmax in saliva at 5 mg | Saliva, not plasma or serum. |
| PMID 35918587 | the endogenous profile; AUC to the last sample | No dose given; the AUC’s end time is not in the record. |
| PMID 37371523 | Cmax after a 100 mg intravenous bolus; Cmax “in the 1 to 3 mg range” | Intravenous; a range of doses, not one. |
| PMID 38293391 | urinary excretion; agomelatine | Not a compared summary; another compound. |
| PMID 36752886 | protein binding; agomelatine’s half-life | Not a compared summary; another compound. |
| PMID 40722923 | bioavailability | Not a compared summary. |
| PMID 41126740 | bioavailability, 1–74% | Not a compared summary. |