Working with real-world data (RWD) to support research is inherently challenging as it is collected primarily for patient care. There are significant variations in how the data is collected and managed between NHS Hospitals, even if the same Electronic Patient Record (EPR) system is used. Without robust data standardisation, RWD can’t be used reliably for large real-world evidence (RWE) studies. 

The process of data standardisation involves two key steps: first, aligning the original NHS Hospital’s data schema with a common data model; and second, mapping each individual medical concept such as specific lab tests or medications to a standardised terminology.

This second step is particularly critical for ensuring the quality and consistency of RWE studies, but it is also time consuming, typically requiring clinical experts to manually map each concept. One widely adopted standard is the OMOP (Observational Medical Outcomes Partnership) common data model, which leverages standardised vocabularies such as LOINC (Logical Observation Identifiers Names and Codes) and SNOMED (Systematised Nomenclature of Medicine).

The complexity increases further when data is sourced from multiple trusts, each with its own documentation style. For instance, in a recent RWE study, we encountered four different medical terms from different NHS Hospital, all referring to the same standard concept: Serum alanine aminotransferase level.As illustrated in this example, some of the original medical concepts are very close to the common concept such as “Alanine Transaminase”, however others require expert knowledge such as “UELFT BONE_XHIT_ALT”.

To streamline the data standardisation process, researchers at Arcturis have developed ArcMAP, an advanced machine learning-based tool designed specifically for mapping medical concepts to standard terminologies. ArcMap transforms the traditional approach to data standardisation by offering several key advantages:

• High Efficiency: ArcMAP intelligently analyses original medical concepts from individual NHS Hospitals and automatically suggests the most likely standardised equivalents. Clinical experts are only required to review and confirm these suggestions, significantly reducing the manual effort typically involved in the process.
• Scalability: ArcMAP continuously learns and improves with each reviewed mapping. As new data is introduced—for example, from an unfamiliar NHS Hospital, ArcMAP draws on its growing understanding of medical language and documentation styles to make more accurate predictions.
• Traceability: Every mapping action is recorded in an internal database, capturing details such as the reviewer, timestamp, and any subsequent modifications. This comprehensive audit trail ensures full transparency and supports regulatory compliance, making the data suitable for submissions and longitudinal studies.