Retrieval-Augmented Interpretable Learning: Towards Task-Specific Zero-Shot Models in Healthcare

2026-07-19

AI Summary

This work addresses the challenges of long-tailed data distributions, frequent emergence of new tasks, and scarce annotations in clinical settings, which undermine model interpretability and reliability. To tackle these issues, the authors propose RAIL, a novel framework that integrates retrieval-augmented inference with interpretable learning. RAIL leverages natural language task descriptions and memory of historical tasks to generate clinical prediction models capable of feature-level explanations under zero-shot or few-shot (2–4 examples) conditions. By combining probabilistic meta-learning, coefficient-space transfer, and uncertainty quantification, RAIL enables multi-level uncertainty estimation and facilitates human-in-the-loop validation. Experiments demonstrate that RAIL achieves 73.4% accuracy in zero-shot settings and maintains 73.2% with minimal supervision—significantly outperforming conventional supervised baselines—while providing transparent retrieval rationales and diagnostic explanations.

Abstract

We introduce Retrieval-Augmented Interpretable Learning (RAIL), a probabilistic meta-learning framework for zero-shot generation of task-specific interpretable models that synthesizes coefficient-space structure from natural-language task descriptions and a memory of previously learned task-specific predictors. RAIL retrieves related source tasks, transfers structure through coefficient space, and generates a new predictor in the original diagnostic-feature space, enabling zero-shot and few-shot clinical procedure prediction with feature-level explanations. Its probabilistic formulation provides uncertainty over retrieval, model coefficients, and predictions, supporting reliability-aware deployment: uncertain predictions or unstable explanations can be flagged for additional clinical review rather than treated as automatic decisions. This makes RAIL particularly suited for healthcare settings, where prediction tasks are highly long-tailed, new clinical targets arise frequently, and models must remain inspectable, uncertainty-aware, and compatible with human oversight. Across long-tailed clinical procedure prediction tasks, RAIL maintains reliable performance across data-availability regimes: it achieves 73.4% accuracy in the held-out zero-shot settings, where no supervised task-specific model can be trained, and remains near 73.2% accuracy in the extreme few-shot regime with only 2-4 examples, where supervised task-specific models perform close to chance. RAIL further benefits from clinically informed task representations and yields retrieval, uncertainty, and coefficient-level diagnostics that make model behavior more transparent. These results suggest a path toward scalable clinical prediction systems that can adapt to new tasks while preserving interpretability and reliability.

Problem

• zero-shot learning
• interpretable models
• clinical prediction
• long-tailed tasks
• healthcare AI

Innovation

• Retrieval-Augmented Learning
• Zero-Shot Learning
• Interpretable Models
• Probabilistic Meta-Learning
• Clinical Prediction