Comparing Knowledge Source Integration Methods for Optimizing Healthcare Knowledge Fusion in Rescue Operation

TL;DR

This paper addresses the challenge of integrating diverse medical knowledge for time-sensitive rescue operations by proposing two conceptual fusion models grounded in knowledge graphs and Bayesian networks. The first model attaches Bayesian-inferred probabilities as edge weights in a main KG, enabling context-aware, probabilistic decision support, while the second model proactively enlarges the KG by contextually integrating multiple external sources through ontology-driven alignment. Using the KIRETT rescue-operations KG as a foundation, the authors illustrate implementation strategies, including a BN-based example for acute coronary syndrome and external knowledge fusion with sources like BPR_SAA_2023, to expand treatment options. The work advances practical decision-support in healthcare by offering structured methods to fuse numerical inferences and contextual knowledge, with potential to improve real-time, patient-specific rescue outcomes. The significance lies in providing two complementary pathways to enrich medical knowledge graphs with probabilistic reasoning and broader sources, supporting more informed, context-sensitive clinical decisions in critical care settings.

Abstract

In the field of medicine and healthcare, the utilization of medical expertise, based on medical knowledge combined with patients' health information is a life-critical challenge for patients and health professionals. The within-laying complexity and variety form the need for a united approach to gather, analyze, and utilize existing knowledge of medical treatments, and medical operations to provide the ability to present knowledge for the means of accurate patient-driven decision-making. One way to achieve this is the fusion of multiple knowledge sources in healthcare. It provides health professionals the opportunity to select from multiple contextual aligned knowledge sources which enables the support for critical decisions. This paper presents multiple conceptual models for knowledge fusion in the field of medicine, based on a knowledge graph structure. It will evaluate, how knowledge fusion can be enabled and presents how to integrate various knowledge sources into the knowledge graph for rescue operations.

Figures (5)

• Figure 1: Contextual Knowledge Fusion: KG I and KG II presents two knowledge graphs, which are contextually interconnected with each other. KG III presents an external knowledge source, which is based on a Bayesian network. In this, a within-calculated Bayesian interference is presented, which has an impact on the contextually aligned knowledge graphs (KG I and KG II). Such inference can be furthermore integrated on the edges of the knowledge graphs to prepare probabilities for assistive decision-making.
• Figure 2: Integration of Bayesian network into an acyclic knowledge graph. Presented are external knowledge sources as a Bayesian Network. The graph (yellow, orange, blue) describes different treatment paths, calculated inferences for each path. Those inferences are integrated as weight into the relations of the second knowledge graph, to support in decision, like treatments. The Bayesian inference supports in decision-making for health professionals based on different external graph sources and historical data.
• Figure 3: Knowledge fusion through contextual node-correlation:$KG_{1}$ describes the knowledge graph, which covers a specific domain-knowledge. $KG_{n}$ describes n-many external knowledge sources (a knowledge graph, a manual, an instruction guideline) which have a contextual correlation with $KG_{1}$. The integrated KG (right) describes the resulting contextual combined knowledge graph with additional nodes for knowledge expansion on the $KG_{1}$.
• Figure 4: Transfer of Bayesian inference into the knowledge graph for rescue operation. On the left side is the treatment path of the "Acute coronary syndrome" presented and on the right side the Bayesian Network of an external knowledge source is visualized. The calculated Bayesian inference can be integrated into the treatment path, to provide probabilities based on historical data for rescue operators.
• Figure 5: Knowledge fusion in the treatment path of the acute coronary syndrome. The depicted flow chart is the treatment path for acute coronary syndrome, extracted from the treatmentpdf manual for rescue operators. (Blue circle) describe the treatment path for acute coronary syndrome, constructed by medical experts treatmentpdf and designed as a graph zenkert_kirett_2022. (Yellow circle) presents the knowledge fusion approach based on contextual alignment. The knowledge integrated was extracted from BPR_SAA_2023. The orange arrows describe an exemplary recommendation path, based on tacit knowledge, if external knowledge sources are integrated.