Progression of risk in heart failure using dynamic risk modelling

Kazmi, Syed

Abstract

Heart failure (HF) is a prevalent condition affecting a significant number of individuals in the UK, leading to substantial healthcare utilisation and adverse outcomes. Despite advancements in treatment and management, the prognosis for hospitalised HF patients remains poor, with a one-year mortality rate of 40%. Improving risk modelling and predictive assessment is crucial for enhancing patient outcomes and reducing healthcare burden.
This thesis aims to analyse HF progression, improve data handling for risk analysis, and develop machine learning (ML) models for tracking health status changes and predicting risks over the course of the disease. To do this a dynamic risk modelling approach was developed and used. This started with the use of a Naïve modelling process with standard ML methodology to the use of Markov chains (MC) and multistate modelling (MCM) with modified MC. The models incorporated dual temporal perspective to investigate the progression of risk which comprises of both short-term and long-term prediction, enabling more accurate forecasting of patient outcomes across varying timeframes.
It was found that the MSM methods can predict a) hospitalisation and mortality both at population level and individual level b) can also determine the number of time visit are made to both [Hosp] and [OPD] before patient died. An expected finding of this thesis is that the progression of HF is linear and not non-linear as it has been assumed. At the same, the modelling in this thesis contributed valuable insights into the progression of risk in HF and underscored the importance of dynamic risk modelling for prognostic assessment. Recommendations for future research include further validation and refinement of the model to enhance predictive accuracy and clinical utility.

Citation

Kazmi, S. (2024). Progression of risk in heart failure using dynamic risk modelling. (Thesis). University of Hull. https://hull-repository.worktribe.com/output/5086141

• Thesis Type: Thesis
• Deposit Date: Mar 20, 2025
• Publicly Available Date: Apr 8, 2025
• Keywords: Computer science
• Public URL: https://hull-repository.worktribe.com/output/5086141
• Additional Information: School of Computer Science; University of Hull
• Award Date: Nov 13, 2024