Preserve
the morning peak (CAR).
A high-amplitude cortisol awakening response is healthy. The controller must not flatten it — a flat rhythm is itself pathological.
Cortisol · circadian controller
One multi-modal candidate to hold cortisol in its healthy band.
Cortisol is not a variable you minimize — you need the morning peak and a high-amplitude rhythm. Disease is dysregulation of the trajectory: a raised nocturnal nadir, a blunted slope, exaggerated reactive spikes. So the target is band control, not suppression. DiscoveryLab composes a single multi-modal candidate across the HPA axis — a CRF1-antagonist peptide, an 11β-HSD1 inhibitor, a selective GR modulator, and a psychobiotic program — and certifies whether the combined trajectory stays inside the healthy circadian band, measured continuously by a wearable free-cortisol sensor, where every single lever under-controls or over-suppresses.
FIG · DiscoveryLab workspace pattern used by the cortisol vertical
The controlled variable
Cortisol is a setpoint-band problem — which is exactly what the controller layer is for.
Treating cortisol as 'lower is better' is how single agents fail: a synthesis inhibitor cranked up flattens the rhythm or pushes the axis into insufficiency. The right objective is to hold the free-cortisol trajectory inside a healthy band across the day — high-amplitude morning peak, low nocturnal trough, controlled reactive excursions. That is a control target with a peak and a trough, not a number to drive to zero.
Lower
the elevated nocturnal nadir.
The disease signature is a raised evening/night trough and a blunted diurnal slope. The controller pulls the trough down toward the healthy floor.
Damp
pathological reactive spikes.
Stress- and meal-driven excursions are attenuated centrally — without abolishing a normal acute response.
Never
over-suppress below the ceiling.
Adrenal insufficiency, AM cortisol below the insufficiency floor, loss of circadian amplitude, immunosuppression — the bound the combination must provably stay above.
The wearable endpoint
Continuous free-cortisol sensing is the discriminator the proof layer needs.
Our verified-controller predicate requires an orthogonally-measurable endpoint checked at both the peak and the trough of the dosing cycle. For cortisol the dosing cycle is the circadian rhythm itself — and historically the endpoint was a few blood draws or saliva spot-checks. First-in-human continuous, multi-day, free-cortisol wearables (e.g. Adaptyx Biosciences, presented at ADA 2026) change that: a time-resolved trajectory non-invasively, day and night.
Defines the band
A continuous trajectory turns 'healthy cortisol' from a single number into an observed peak, slope, and trough you can fit a target band to.
Verifies in-band control
The controller claim — stays inside the band across the cycle — becomes checkable against real day-and-night data, not an idealized steady state.
Binds the kill-criterion
The pre-registered readout is the wearable signal: super-additive band restoration over the best single lever, or it is falsified.
This is the rare case where the instrument for the controlled variable now exists continuously and non-invasively — turning cortisol regulation from an endpoint-poor target into an unusually clean controller problem.
Three paths
One axis — the HPA loop — reached three different ways.
No single modality reaches every node of the HPA axis. DiscoveryLab runs all three, each at its own node and timescale, each with its own honest claim level, and lets the verification layer decide what combines into a single candidate.
Path 1 · direct
discoveryPeptides at the HPA control nodes
The hypothalamic–pituitary–adrenal axis is run by peptide signalling — CRH, ACTH, urocortins. Peptides antagonise those receptors with sequence-level specificity, trimming the central drive and the adrenal output ceiling without abolishing the rhythm.
CRF1 / CRHR1 antagonistMC2R / ACTH-receptor capUrocortin / CRF-BPfast · reactive spikes
Calibrated reference modality. Signed mechanism attribution + machine-checked Lean receipts. Carries the discovery claim in the combination below.
Path 2 · enzymatic
rank-onlySmall molecules over cortisol metabolism
The validated drug levers act on tissue-level cortisol, not adrenal output. An 11β-HSD1 inhibitor cuts the regeneration of active cortisol in liver, adipose, and brain; a selective GR modulator caps glucocorticoid signalling at end organs during excess.
11β-HSD1 inhibitorSelective GR modulatortissue · intermediatespares circulating rhythm
ChemCheck v0: PAINS/Brenk + Lipinski/Veber developability triage plus occupancy from supplied affinity + free concentration. Rank-only until calibrated on measured outcomes.
Path 3 · indirection
rank-onlyMicrobiome-mediated HPA tone
The gut–brain axis sets the HPA setpoint. Steer the ecosystem so vagal and metabolite routes — butyrate/SCFAs, microbial tryptophan and GABA — lower baseline tone and stress reactivity, the slow floor the fast levers ride on top of.
Psychobiotic strainsButyrate → vagal toneTrp / GABA routesslow · baseline + nadir
Ecological dependency graph (helper → producer → metabolite → vagus/host receptor → HPA tone). Measurement-first plans only; no colonization, flux, or efficacy claim until calibrated.
The candidate
cortisol_band_controller_v0 — four orthogonal levers, one composition.
Each lever hits a different node of the same variable on a different timescale — so they compose rather than stack. The microbiome program sets the slow floor; the peptide caps fast spikes; the 11β-HSD1 inhibitor trims tissue-level cortisol; the GR modulator bounds end-organ signalling. The point is control, not breadth.
Proposed composition · HPA circadian bandcertified · Lean receipt
01
Microbiome
rank-onlypsychobiotic + SCFA program
Baseline tone & nocturnal nadir
Lowers the chronic floor and stress-reactivity setpoint via vagal and metabolite routes — the slow base the fast levers sit on.
slow · chronic
02
Peptide
discoverycrf1_antagonist_peptide
Central drive (hypothalamus/pituitary)
Caps fast, stress-reactive cortisol excursions centrally — without abolishing the needed morning cortisol awakening response.
fast · reactive
03
Small molecule
rank-only11β-HSD1 inhibitor
Tissue cortisol regeneration
Trims active intracellular cortisol exactly where the metabolic and cardiovascular damage happens, sparing the circulating circadian rhythm.
intermediate
04
Small molecule
rank-onlyselective GR modulator
End-organ glucocorticoid signalling
A safety-bounded ceiling at target tissues during excess — caps signalling without zeroing the receptor.
intermediate
Why every monotherapy under- or over-controls: an 11β-HSD1 inhibitor alone fixes tissue amplification but triggers compensatory ACTH rise and ignores reactive spikes; a CRF1 peptide alone blunts spikes but risks flattening the morning peak; the microbiome program alone moves baseline slowly and weakly; any synthesis inhibitor cranked up alone over-suppresses into a flat rhythm. That gap between every single lever and adequate control is exactly what the combination is built to close.
Controller & kill-criterion
Held in-band, bounded below the ceiling, and falsifiable on the wearable.
Treated as a controller, the question is not 'is each agent good?' but 'does the combined intervention keep the free-cortisol trajectory inside its healthy band across the full circadian cycle, where each single lever provably under-controls?' That predicate is now machine-checked: the Lean receipt lyapunov_cortisol_combination proves stability at peak AND trough, in-band (K 0.116 / 0.108), below the over-suppression ceiling, and strictly better than any monotherapy (mono K 0.064 < 0.100). It remains an in-silico certificate over documented surrogate constants — not a validated therapy.
Certified · cortisol now carries its own CI-checked Lean receiptLean receipt
The combination-controller layer certifies multi-modal controllers with CI-recompiled Lean receipts — and the cortisol band controller now has one of its own, lyapunov_cortisol_combination, alongside the glucose axis it sits directly upstream of. The proof recompiles under lake build on every commit.
| Axis | K peak / trough | Robust* | Lean receipt |
|---|---|---|---|
Cortisol circadian band HPA · this program · peak + trough proven | 0.116 / 0.108 | 74% | lyapunov_cortisol_combination |
Glucose lowering metabolism · T2D triple · cortisol is upstream | 0.126 / 0.108 | 74% | lyapunov_glucose_combination |
IL-6 trans-signaling inflammation | 0.131 / 0.109 | 79% | lyapunov_il6_combination |
C5a / C5aR1 inflammation | 0.112 / 0.106 | 67% | lyapunov_c5a_combination |
Band kMin–kMax = 0.10–0.16; every monotherapy under-controls. *Robust = share of an ±15% potency × PK-trough grid (81 worlds, 9×9) in which the combination stays a PK-robust controller. Cortisol holds in 74% of worlds and tolerates a uniform −10% potency drop at nominal PK; the dominant edge failure is the dosing-cycle trough — reproduce with peptiter-reproduce-combination.
Over-suppression ceiling (the Safe bound)
The combination must provably stay above adrenal insufficiency: AM cortisol below the insufficiency floor, collapse of circadian amplitude, a blunted diurnal slope, or immunosuppression. A flat rhythm is a failure mode, not a success — which is why no single lever is allowed to drive the variable to zero.
Named kill-criterion
Falsified if, on continuous wearable free-cortisol, the candidate fails to produce super-additive restoration of circadian amplitude / nocturnal-nadir reduction over the best single lever — or if any arm trips the over-suppression ceiling.
Honest scope This is a machine-checked design with documented surrogate constants — in-silico, not a clinical claim. The Lean receipt proves the controller predicate holds given those surrogates; the peptide lever carries the discovery claim, the small-molecule and microbiome levers are rank-only, and the only bridge to the clinic is one continuous-wearable factorial with the kill-criterion named in advance.