Integrated Quantitative Systems Pharmacology model of dry eye disease, linking tear-film hyperosmolarity, meibomian-gland (evaporative) and lacrimal-gland (aqueous-deficient/Sjögren) dysfunction, the innate→adaptive ocular-surface inflammatory cascade (dendritic cell → Th17/IL-17A → MMP-9), epithelial barrier breakdown, and corneal neurosensory dysfunction — with the modern topical pharmacology stack (cyclosporine, lifitegrast, loteprednol, perfluorohexyloctane, varenicline nasal spray, diquafosol) and the clinical endpoints used to track disease (OSDI, TBUT, corneal staining, Schirmer test, tear osmolarity, MMP-9).

Dry eye disease (DED) affects 5-50% of adults worldwide (TFOS DEWS II) and is defined as a multifactorial disease of the ocular surface characterized by a loss of tear-film homeostasis, accompanied by ocular symptoms in which tear-film instability and hyperosmolarity, ocular surface inflammation and damage, and neurosensory abnormalities play etiological roles. Two overlapping phenotypes converge on the same vicious cycle: evaporative DED (meibomian gland dysfunction → lipid-layer deficiency → increased evaporation) and aqueous-deficient DED (lacrimal-gland dysfunction, often Sjögren- associated → reduced tear production). Either route raises tear-film osmolarity, which activates MAPK/NF-κB/NLRP3 signaling in ocular-surface epithelium, triggering an innate inflammatory cascade (IL-1β, TNF-α, MMP-9) that matures dendritic cells and primes an adaptive Th17/Th1 response; effector cytokines (IL-17A, IFN-γ) home back to the ocular surface, causing goblet-cell loss, epithelial barrier breakdown, and further tear-film destabilization — closing a self-perpetuating loop. Chronic disease also damages corneal sub-basal nerves, producing peripheral/central sensitization and neuropathic ocular pain that can persist even as signs improve. Modern therapy targets specific nodes of this cycle: cyclosporine A and lifitegrast dampen the DC/Th17 axis; loteprednol provides rapid but time-limited broad anti-inflammatory pulse therapy; perfluorohexyloctane is a non-drug evaporation barrier; varenicline nasal spray stimulates reflex tearing via the trigeminal parasympathetic pathway; and diquafosol (P2Y2 agonist, approved in Japan/Korea) stimulates mucin and aqueous secretion directly.

1. Epidemiologic & environmental drivers (age, sex/androgen, screen time, CL wear, refractive surgery, autoimmune disease, BAK preservative toxicity)
2. Trilaminar tear-film homeostasis (lipid/aqueous/mucin layers, TBUT, evaporation)
3. Meibomian gland dysfunction (evaporative DED — keratinization, obstruction, gland dropout)
4. Lacrimal gland dysfunction (aqueous-deficient DED — Sjögren infiltrate, acinar apoptosis, reflex-arc impairment)
5. Hyperosmolar stress signaling (MAPK/NF-κB/NLRP3 inflammasome, ROS)
6. Innate inflammatory cascade (IL-1β/TNF-α/IL-6/IL-8/MMP-9, neutrophil recruitment, DC activation)
7. Adaptive immunity — self-perpetuating vicious cycle (DC maturation → Th17/Th1 → IL-17A/IFN-γ → ocular homing → chronic inflammation feed-forward)
8. Corneal neurosensory dysfunction (TRPM8 cold receptors, nerve loss, peripheral/central sensitization, neuropathic pain, photophobia)
9. Ocular-surface epithelial damage (goblet-cell loss, squamous metaplasia, apoptosis, barrier dysfunction, filamentary keratitis)
10. Topical & systemic drug PK/PD (cyclosporine, lifitegrast, loteprednol, perfluorohexyloctane, varenicline, diquafosol, punctal plugs, autologous serum, omega-3, thermal pulsation/IPL)
11. Clinical endpoints & biomarkers (OSDI, eye-dryness VAS, TBUT, Schirmer, corneal/conjunctival staining, tear osmolarity, MMP-9 POC, meibography, NEI-VFQ-25)

• Local/topical drug exposure (8 compartments) — cyclosporine A (0.05%/0.09%), lifitegrast 5%, loteprednol 0.25% (+ cumulative-exposure tracker for IOP-risk), perfluorohexyloctane, varenicline nasal spray (nasal depot + trigeminal-reflex effect compartment), diquafosol 3%. Topical ophthalmic PK is dominated by tear-turnover on a minutes scale, so each drug is represented as a day-resolution surrogate "sustained local exposure" compartment rather than a literal plasma PK model.
• Tear-film & ocular-surface disease network (14 compartments) — tear osmolarity, lipid-layer quality, aqueous production drive, mucin/goblet density, dendritic-cell activation, IL-17A/Th17 tone, MMP-9, epithelial barrier integrity, corneal nerve density, neuropathic pain, TBUT, Schirmer, corneal staining, OSDI — the last four double as clinical-endpoint readouts.

1. Patient profile — MGD/ADDE severity sliders, Sjögren flag, refractive-surgery flag, baseline OSDI.
2. Drug exposure — fractional drug-effect trajectories for all six agents.
3. Tear-film / PD — tear osmolarity, lipid-layer quality, aqueous production, mucin/goblet density.
4. Inflammatory cascade — dendritic-cell activation, IL-17A tone, MMP-9, epithelial barrier integrity.
5. Clinical endpoints — TBUT, Schirmer test, corneal staining, OSDI symptom score.
6. Neurosensory — corneal nerve density and neuropathic ocular pain score.
7. Scenario comparison — runs all 9 built-in scenarios with the chosen patient profile.
8. Biomarkers / safety — loteprednol cumulative-exposure IOP-risk score, evaporation-rate trajectory, endpoint summary table.

# 1) Render the mechanistic map
dot -Tsvg ded_qsp_model.dot -o ded_qsp_model.svg
dot -Tpng -Gdpi=150 ded_qsp_model.dot -o ded_qsp_model.png

# 2) Simulate scenarios in R
install.packages(c("mrgsolve","dplyr","tidyr","ggplot2","shiny",
                   "shinydashboard","DT"))
library(mrgsolve)
mod <- mread_cache("ded_mrgsolve_model.R")
out <- mrgsim(mod, ev(amt=1.0, cmt="CSA_TF", ii=0.5, addl=168), end=84)
plot(out, c("OSDI_score","TBUT_s","Corneal_staining","Schirmer_mm"))

# 3) Launch the dashboard
shiny::runApp("ded_shiny_app.R")

• Designed for research, education, and hypothesis generation — not clinical decision support.
• Topical ocular drug "exposure" compartments are pragmatic day-resolution surrogates for sustained local tissue concentration, not literal tear-film or aqueous-humor pharmacokinetics (which occur on a minutes-to-hours scale dominated by tear turnover).
• Perfluorohexyloctane and diquafosol are modeled with no direct anti-inflammatory arm, consistent with their non-immunomodulatory mechanisms (physical evaporation barrier and P2Y2 secretagogue, respectively).
• Corneal nerve density and neuropathic pain evolve on a much slower (months-scale) timescale than the inflammatory/tear-film compartments; the 9 built-in scenarios (12-week default horizon) will show only partial recovery of these compartments.
• Loteprednol's IOP-risk score is a simplified linear function of cumulative exposure intended to flag the real-world need to limit corticosteroid courses to 2-4 weeks — it is not a validated IOP-elevation model.

Inherits the repository LICENSE.