Booth, John Hugh; (2025) Determining patient intensity of care using routinely collected Electronic Patient Record (EPR) data and graph analytics. Doctoral thesis (Ph.D), UCL (University College London).

Text Booth_10214926_Thesis.pdf Access restricted to UCL open access staff until 1 April 2026. Download (21MB)

Abstract

The intensity of care supplied to a patient during a hospital admission is a complex measure and difficult to interrogate using existing methods of analysis on Electronic Patient Record (EPR) data. The objective of this study was to determine the feasibility of applying graph analytics to EPR data to analyse the interconnections between data points and whether these measures could be used to reflect different levels of intensity of care. Temporal graphs representing the daily interactions between patients and healthcare professionals (HCPs) were created and shown by visual analysis to represent daily changes in intensity of care. Categorisation of the temporal graphs using k-means clustering with the graph’s metadata as features was carried out. This categorisation was then used to categorise post-operative care in patients who had undergone a renal transplant using each day of post-operative care as the feature and the temporal graph categorisation as the value. A reproducible analytical pipeline (RAP) was developed that generated and categorised temporal graphs. Several different EPR datasets were selected and a combination of HCP interactions, clinical note production and bed-side administrations were identified as a more complete representation of a patient’s intensity of care. To establish a wider application for the RAP, EPR data for a single ward over a period of a month was extracted and three sets of temporal graphs, individual patients, the complete ward, using 8-hour periods rather than 24-hours, and individual HCPs, were generated and categorised. This process proved to be a novel method of clearly demonstrating the variations in intensity of patient care in different situations. This thesis concludes that the application of temporal graph analysis on routinely collected EPR data can provide novel insights into patient care, patient outcomes, and hospital operations. Although not a standalone application these techniques adds nuance to traditional analysis both visually and objectively. By employing RAP concepts, these techniques will be able to be deployed throughout the NHS.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as XMLJSONCSV

Archive Staff Only

View Item