Learning instance-specific counterfactual models for continuous treatments using hypernetworks
Roger Pros, Jordi Vitrià · 2026
Read Learning instance-specific counterfactual models for continuous treatments using hypernetworks by Roger Pros, Jordi Vitrià (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13194395.txt.
Description
Roger Pros and Jordi Vitrià tackle in 2026 a causal question with no off switch: what would happen to this patient at every dose of a treatment, not just treated or not. Most machine learning for treatment effects handles a yes or no intervention. Real medicine titrates: drug doses, exposure levels, and therapy intensities live on a dial.
Their architecture uses a hypernetwork, a network whose job is to manufacture another network's weights. For each person and each dose level, it generates a predictor tuned to that exact scenario, keeping the dose's causal role explicit. On synthetic and semi synthetic data the method estimated dose response curves more precisely than existing rivals. Hard cases are its home turf: many confounders and bending, nonlinear responses. Tools of this kind aim at personalized dosing, where the question is never whether to treat but how much.
Catalog record
| CREATOR | Roger Pros, Jordi Vitrià |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |