EURO 2025 Leeds
Abstract Submission

1671. A Two-Stage Machine Learning and Evidential Reasoning Framework for Trustworthy Decision Support

Invited abstract in session TA-28: AI and Machine Learning for Decision Support, stream Decision Support Systems.

Tuesday, 8:30-10:00
Room: Maurice Keyworth 1.03

Authors (first author is the speaker)

1. Junhong Guo
Alliance Manchester Business School, The University of Manchester
2. Dong-Ling Xu
Alliance Manchester Business School, The University of Manchester
3. Jian-Bo Yang
Alliance Manchester Business School, The University of Manchester

Abstract

Balancing predictive performance and interpretability is a critical challenge in decision support systems, particularly in high-stake domains such as medical diagnosis and financial risk assessment, where both accuracy and transparency are essential. Traditional interpretable models struggle to capture latent relationships among complex data, while high-performance machine learning (ML) models lack transparency, limiting their applicability in high- stake decision-making. To address this issue, we propose a Two-stage Machine Learning-Evidential Reasoning (TMLER) framework, which integrates ML and Evidential Reasoning (ER) to strike an optimal balance between accuracy and interpretability. In the first stage, ML methods extract high-level, human-recognizable features, such as medical risk scores or imaging-derived indicators, from multimodal data, replacing manual preliminary assessments. In the second stage, these extracted features are systematically fused within the transparent ER framework to combine evidence and handle uncertainty, generating structured and traceable decision support. Experimental validation on real-world medical datasets demonstrates that TMLER provides structured and explainable decision support without compromising predictive performance, offering a viable solution for trustworthy AI-driven decision-making.

Keywords

Status: accepted

Back to the list of papers