Constrained conditional LLM-based imputation via iterative chained updates

Andrea F. Abate

doi:10.1016/j.patrec.2026.06.015

Constrained conditional LLM-based imputation via iterative chained updates

2026-06-19

0

OA

AI

DOI：10.1016/j.patrec.2026.06.015

Original

Original request for help

Save

Abstract

En 中文

• Novel chained LLM-based framework for tabular data imputation. • Clinical LLM backbone selected via intrinsic validation protocol. • LoRA-based backbone adaptation for diabetic binary classification. • Controlled experiments under different missingness mechanisms. • Language influences LLM-based structured imputation behavior.

Keywords:

Large language models

Data imputation

Iterative chained updates

Conditional inference

Biomedical data

Constrained conditional LLM-based imputation via iterative chained updates

Abstract

Journal

Organization

Cited Papers