Identifying Subgroups in Heart Failure Patients with Multimorbidity by Clustering Analysis

November 26, 2021

This dissertation proposes a workflow for the identification and characterisation of clinically significant Heart Failure (HF) patient subgroups with multimorbidity using data from Electronic Health Records. Multimorbidity, defined as the co-occurrence of two or more chronic conditions, represents a heavy burden for healthcare systems. However, there is still a lack of knowledge about patients with multimorbidity, the most common disease interactions, risk factors and treatment response. This is particularly relevant in complex and heterogeneous conditions such as HF. Using clinical data from Hospital da Luz Lisboa, we conducted a clustering analysis of HF patients and characterised the clusters in terms of demographics, comorbidities, laboratory values, medical prescriptions and medical appointments. This was followed by a survival analysis to identify associations between clusters and outcomes such as hospitalisation and emergency admission. The workflow identified four distinct clusters with significant differences in the clinical variables analysed. The survival analysis showed differential associations for hospitalisation and emergency admission risks between clusters. Moreover, the risk associations were in agreement with the cluster profiles. These results evidence a high degree of disease heterogeneity within HF patients and how an improved characterisation of patient subgroups can be relevant for clinical risk stratification.

Authors in the community:

• 

  C

  Catarina Patrício Mendes Martins

  ist186766

Supervisors of this institution:

• 

  M

  Mário Jorge Costa Gaspar da Silva

  ist124073

industrial-biotechnology - Industrial Biotechnology

eng - English

Instituto Superior Técnico