This post explains our published research: A Bistable ODE Model Predicts Durable Off-Treatment Resolution of Chronic Plaque Psoriasis via Tissue-Resident Memory T Cell Depletion (Rai, 2026, DOI: 10.5281/zenodo.19499470).
About 125 million people live with plaque psoriasis, and nearly all of them know the same cycle: a treatment works, the skin clears, the treatment stops — and the plaques come back, usually in exactly the same spots. Modern biologic drugs are genuinely good at suppressing the disease. What they are not is curative. Why?
The answer, increasingly clear from immunology, is a cell type with a bureaucratic name: the tissue-resident memory T cell, or TRM. These are immune cells that take up permanent residence in previously inflamed skin and sit there for years, quietly holding the memory of the disease. Biologics silence the inflammatory noise, but they do not evict these tenants — so the moment the drug stops, the tenants restart the fire. Same spots, every time.
Our study built a mathematical model of this whole circuit — twenty interacting variables covering the immune cells, the skin cells, and the mechanical stress of the thickened plaque, calibrated against published single-cell and pharmacokinetic data. The model is bistable, which is the mathematical way of saying the skin has two stable states, healthy and psoriatic, like a light switch. Suppressing symptoms holds the switch halfway; a cure means flipping it fully back and having it stay.
Then the model screened 5,757 combinations of 21 agents, looking not for the strongest suppression but for the combination that empties the TRM reservoir — the only exit from the loop. The winning protocol has five inexpensive components used together for a defined period: narrowband UVB phototherapy three times a week, a topical andrographolide gel, oral berberine, low-dose weekly methotrexate, and very-low-dose weekly rapamycin. Nothing exotic; the novelty is the combination and the target. The model predicts an 85–87% rate of permanent clearance at twelve months after stopping everything, at a monthly cost around 345–405 US dollars — a fraction of what biologics cost.
Now the honesty paragraph, because it matters more here than anywhere. This result lives inside a computer. No patient has received this protocol in a trial. Two of its five components — methotrexate and rapamycin — are serious prescription drugs with real risks that need medical supervision and monitoring, and “low dose” is not “safe by default.” A model prediction of this kind is a testable hypothesis and an argument for a clinical trial; it is not something to assemble at home. If you have psoriasis, the actionable content of this research is a question for your dermatologist: is my treatment plan only suppressing, or is anyone thinking about the memory cells? — plus the knowledge that curative-intent research on this disease exists and is accelerating.
The full model, protocol and every equation are open-access at zenodo.org/records/19499470.