Prior-Fitted Functional Flow: In-Context Generative Models for Pharmacokinetics

TL;DR

The paper presents Prior-Fitted Functional Flows, a generative model for pharmacokinetics that enables zero-shot population synthesis and individual forecasting with calibrated uncertainty.

Contribution

It introduces a novel approach to learn functional vector fields conditioned on sparse data, improving pharmacokinetic predictions without manual tuning.

Findings

Achieves state-of-the-art accuracy on real-world datasets.

Enables coherent virtual cohort generation.

Provides calibrated uncertainty in forecasts.

Abstract

We introduce Prior-Fitted Functional Flows, a generative foundation model for pharmacokinetics that enables zero-shot population synthesis and individual forecasting without manual parameter tuning. We learn functional vector fields, explicitly conditioned on the sparse, irregular data of an entire study population. This enables the generation of coherent virtual cohorts as well as forecasting of partially observed patient trajectories with calibrated uncertainty. We construct a new open-access literature corpus to inform our priors, and demonstrate state-of-the-art predictive accuracy on extensive real-world datasets.