From the workbench

Blog

Methodology notes, replication memos, and short pieces on the mechanics of modelling human physiology. Citation-heavy by design.

Three concentration thresholds for ivermectin (immune, PAK1, cytotoxic) plotted against achievable plasma and tissue levels at oral doses, with mebendazole tubulin EC50 overlaid
Mechanistic Modelling May 28, 2026

Repurposed antiparasitics in oncology: what the mechanism predicts

Ivermectin and mebendazole both have anti-cancer mechanisms that are active at concentrations reachable with oral dosing. The composed body model predicts a real but modest combined effect at 90 days — far below the headline numbers in current observational reports. The trial that would distinguish drug effect from concomitant therapy has not been run.

The hantavirus cascade — organ-by-organ progression from endothelial infection through cytokine self-sustaining loop to multi-organ failure, with each drug's effective treatment window
Mechanistic Modelling May 26, 2026

Hantavirus: the cascade that decouples from the virus

Hantavirus does not kill cells directly. It kills by triggering a cytokine cascade that, once committed, propagates organ by organ — lung capillary leak, vascular shock, marrow suppression, lactic acidosis — independently of the virus that started it. The cascade is what makes the timing window structural, not pharmacological.

Four bone-active drug classes mapped onto the Lemaire-Pivonka coupled bone-remodelling cycle, with each drug's intersection with the tumour-driven vicious cycle of bone metastasis highlighted
Drug Safety May 26, 2026

Bone is a microenvironment. Anabolic drugs change who lives there.

Bisphosphonates and denosumab reduce skeletal-related events in cancer with bone metastases. The anabolic bone drugs — teriparatide, romosozumab — have not been studied in patients with cancer at all. The same physiology that builds bone could plausibly feed an occult metastasis. A coupled Lemaire-Pivonka model lets us ask the question the trials did not.

Three-channel attribution of the GLP-1 cardiovascular benefit — weight-loss-mediated, glucose-mediated, and direct cardiac receptor — with the modelled contribution of each channel to total MACE reduction
Mechanism Attribution May 26, 2026

GLP-1s reduce cardiovascular events. The model says the mechanism is mostly indirect.

SELECT showed semaglutide reduced major adverse cardiovascular events by 20% in non-diabetic patients with established cardiovascular disease. Three plausible mechanisms could explain that signal — direct cardiac receptor effects, glucose-mediated, or downstream of weight loss. A coupled body model says the third channel carries most of the effect — which has direct consequences for the next-generation GLP-1-without-weight-loss molecules now in development.

The asterisk on anti-amyloid Alzheimer drugs — pseudoatrophy versus real neurotoxicity hypotheses diverging past the 18-month trial horizon
Drug Safety May 25, 2026

The asterisk on the anti-amyloid Alzheimer drugs

Lecanemab and donanemab reduce cognitive decline ~25–33% over 18 months. They also shrink the brain — by about half a percent of whole-brain volume in the same window. The pseudoatrophy and the real-neurotoxicity hypotheses make identical 18-month predictions. The data needed to distinguish them does not exist yet.

Ebola case-fatality rate by treatment start day — the 48-hour timing cliff between Day 0 and Day 5+ across mAb114 and REGN-EB3 arms
Trial Replication May 25, 2026

The Ebola drugs work. The timing cliff starts at 48 hours.

The PALM trial established that mAb114 and REGN-EB3 reduce mortality in Ebola. It did not establish how outcomes depend on when treatment starts. A coupled 22-state model, calibrated against all six PALM arms, says the answer is steeper than the field assumes — and the implication is logistical, not pharmacological.

Side-by-side comparison of patient trajectory foundation models (Foresight, ETHOS, DT-GPT) and mechanistic patient digital twins (PBPK, QSP, Living Heart)
Methodology May 25, 2026

Foresight is a remarkable model. It is not a patient digital twin.

Why a generative transformer over electronic health records cannot answer the counterfactual question every clinician actually wants to ask — and what the engineering definition of digital twin requires instead.