Explainable Biomedical Recommendations via Reinforcement Learning Reasoning on Knowledge Graphs

TL;DR

This paper explores the application of a neurosymbolic, knowledge graph-based reinforcement learning approach for transparent and accurate biomedical recommendations, specifically in drug discovery, demonstrating significant performance improvements and biologically relevant explanations.

Contribution

It is the first systematic application of multi-hop reasoning on knowledge graphs to complex biomedical datasets for drug discovery, with comprehensive benchmark comparisons.

Findings

Outperforms baselines by 21.7% on average

Produces biologically relevant explanations

Validates approach on multiple datasets

Abstract

For Artificial Intelligence to have a greater impact in biology and medicine, it is crucial that recommendations are both accurate and transparent. In other domains, a neurosymbolic approach of multi-hop reasoning on knowledge graphs has been shown to produce transparent explanations. However, there is a lack of research applying it to complex biomedical datasets and problems. In this paper, the approach is explored for drug discovery to draw solid conclusions on its applicability. For the first time, we systematically apply it to multiple biomedical datasets and recommendation tasks with fair benchmark comparisons. The approach is found to outperform the best baselines by 21.7% on average whilst producing novel, biologically relevant explanations.