Enhancing the Interpretability of Rule-based Explanations through Information Retrieval

TL;DR

This paper introduces an attribution-based method that enhances the interpretability of rule-based explanations in healthcare AI, specifically for breast cancer lymphedema risk prediction, by analyzing attribute relevance using Information Retrieval metrics.

Contribution

It presents a novel attribution approach that improves interpretability of rule-based AI models in healthcare through statistical attribute relevance analysis.

Findings

Users found the method more interpretable and useful

The approach increased transparency of AI predictions

Enhanced understanding of risk factors in lymphedema prediction

Abstract

The lack of transparency of data-driven Artificial Intelligence techniques limits their interpretability and acceptance into healthcare decision-making processes. We propose an attribution-based approach to improve the interpretability of Explainable AI-based predictions in the specific context of arm lymphedema's risk assessment after lymph nodal radiotherapy in breast cancer. The proposed method performs a statistical analysis of the attributes in the rule-based prediction model using standard metrics from Information Retrieval techniques. This analysis computes the relevance of each attribute to the prediction and provides users with interpretable information about the impact of risk factors. The results of a user study that compared the output generated by the proposed approach with the raw output of the Explainable AI model suggested higher levels of interpretability and usefulness in the context of predicting lymphedema risk.