Digital Twins and Personal Knowledge Graphs in Precision Healthcare: Integrating EHRs, Multi-Omics and Real-Time Patient Data

Abstract

Dynamic, continually updated, and heterogeneous data integration across multiple streams is becoming increasingly important to precision healthcare, where computational models of individual patients are growing more important. Two paradigms have come to the fore for building digital twins: digital twins and Personal knowledge graphs, but they are often treated separately even though they share the same underlying knowledge base, including electronic health records (EHRs), multi-omics profiles, and real-time physiological monitoring. The synthesis of literature from 2020-2025 aimed to map the literature across four areas: (i) the concept and clinical use of health digital twins, ranging from early personalised-medicine framing to formal scoping-review taxonomies; (ii) the use of knowledge-graphs derived from biomedical and electronic-health-record data, as an enrichment layer for patient-specific representations and the early detection of diseases; (iii) computational approaches that allow the integration of multi-omics data as a molecular-enrichment layer both for digital twins and knowledge graphs; and (iv) wearable sensor and remote-monitoring technologies to provide the continuous realtime data needed to keep patient representations up to date. The review reveals that the two literatures of digital twins and personal knowledge graphs are, so far, largely parallel. Still, both are becoming increasingly enriched with patient data from EHRs and omics databases and increasingly informed by sensor data. Both technical literatures draw on similar peer-reviewed evidence from journals indexed by the Institute of Scientific Information, such as npj Digital Medicine, Journal of Personalised Medicine, and Briefings in Bioinformatics. Four comparative tables summarize health digital twin applications and taxonomies, EHRderived knowledge graph applications, multi-omics data integration methods and wearable/real-time (RT) monitoring technologies, and four figures were proposed to visualise an integrated digital twin-knowledge graph patient representation framework, a digital twin maturity taxonomy, an EHR-toknowledge-graph patient embedding pipeline, and a multiomics/wearable (RT) data fusion pipeline. The review suggests that the next generation of precision healthcare infrastructure will likely formalise the integration of digital twin simulation capability with knowledge-graph-based reasoning over a shared, continually updated, multi-modal patient data substrate.