From Knowledge Graphs to Digital Twins: Perspectives on Modeling Patient Outcomes for Health Care Quality Assessment.

Abstract

Medical applications of mathematical modeling, including machine learning models, knowledge graphs, and health digital twins, primarily involve the prediction of patient outcomes. This expert perspective examines how mathematical modeling can contribute to health care quality management. Definitions of procedures, patient outcomes, and quality metrics are provided with a quantitative focus. The emphasis is subsequently placed on 3 categories of patient-centered quality of care, namely, patient safety, procedure accuracy, and procedure efficacy, for which a conceptual and mathematical description is provided. Different levels of modeling tasks essential for managing patient-centered quality of care are identified. This article facilitates a deeper understanding of the topic by assigning relevant publications to these 3 quality categories. Focus is placed on the applicability of graph-based methods, including knowledge graphs and health digital twins, to improve quality management in health care. We have presented a clinical scenario and provided information on methodological limitations, future research directions, and practical implications.