Risk Modelling in Healthcare: A Case Study for Emergency Services

Abstract

Patient experience is a critical aspect of healthcare influenced by various domains related to the providers involved in a patient's hospital visit. While previous studies have explored the relationship between patient experience and provider-related domains, less research has been conducted on the interdependencies between multiple domains and their influence on patient experience. The lack of connection between Bayesian Belief Networks and Patient Experience in Emergency Services appears to be a gap in the existing research. The present study developed a Bayesian Belief Network (BBN) model to address this knowledge gap using patient experience survey data from the UK National Health Service. The model incorporates eight provider-related domains: (i) Arrival at A&E, (ii) Waiting times, (iii) Doctors and nurses, (iv) Care and treatment, (v) Tests, (vi) Environment and facilities, (vii) Leaving the emergency department, and (viii) Respect and dignity in addition to an overall patient experience score. Four learning algorithms were implemented: Bayesian Search (BS), Tree Augmented Naïve Bayesian (TAN), Augmented Naïve Bayesian (ANB), and Naïve Bayesian (NB). Discretization schemes were employed to construct the models. The prediction accuracy for all models were assessed using 10-fold cross-validation. General, the four algorithms demonstrated high accuracy. However, the TAN algorithm with a two-state discretization gets the highest accuracy with 90% . In addition, more analyses were completed on the model, including examinations of sensitivity and scenario-based analyses. The results showed that the most influential domains that led to a high patient experience score were Care and treatment, and Respect and dignity. Also, by combining patient experience data with Bayesian Belief Networks, this study's novel methodology has shed light on the complex interrelationships and dependencies between provider-related domains and their combined influence on the general patient experience in emergency services. Using sophisticated machine learning algorithms and sensitivity analyses, this study provides insights for healthcare decision-makers who wish to improve patient satisfaction and the standard of treatment provided in urgent care environments. The results highlight the critical roles that domains such as arrival at A&E, doctors and nurses, Care and treatment, environment and facilities, and respect and dignity play in patient experience, and they offer practical advice for healthcare organizations that wish to prioritize changes that meet the needs and expectations of their patients.

Date of Award	16 Dec 2023
Original language	American English
Supervisor	Mecit Simsekler (Supervisor)