Personalised patient monitoring in haemodialysis using hierarchical Gaussian processes
The prevalence of end-stage renal failure is 861 per million population in the UK, and these patients undergo three haemodialysis sessions per week, each lasting 4 hours. In addition, patients are at risk of intra-dialytic hypotension, which leads to chronic heart disease and a high incidence of mor...
| Main Authors: | , , , |
|---|---|
| Formato: | Conference item |
| Publicado: |
Institute of Electrical and Electronics Engineers
2017
|
| Summary: | The prevalence of end-stage renal failure is 861 per million population in the UK, and these patients undergo three haemodialysis sessions per week, each lasting 4 hours. In addition, patients are at risk of intra-dialytic hypotension, which leads to chronic heart disease and a high incidence of mortality. Through continuous monitoring of blood pressure during dialysis, we describe the use of Gaussian process regression to model changes of systolic blood pressure over time for each patient. We use Hierarchical Gaussian processes to infer the assumed latent structure of the systolic blood pressure trajectory for each individual patient, to describe their personalised “normal” and “abnormal” physiological patterns. |
|---|