Project Description

We are looking to identify and characterise multimorbid phenotypes with distinct trajectories from EHR data.  Implementation of machine learning algorithms to discover multimorbidity patterns (i.e. how diseases accumulate over time) and approaches to provide more insights into disease co-occurrences for supporting clinicians in taking preventive clinical actions (e.g. early disease diagnosis) for patients at risk.

Project Milestones

  • Patient and public involvement- Research presentation and discussion, Feb 2021
  • Implementation of the developed approach with CALIBER dataset, Feb 2021 – Jun 2021
  • Generating synthetic data sets to demonstrate the utility of the algorithm in idealised conditions, Mar 2021 – Jul 2021
  • Algorithm comparison and implementation with sites, Jul 2021 – Sep 2021
  • Paper write-up, May 2021 – Sep 2021
  • Submitted and work in progress publications are as follows:
    • Discovering multimorbidity patterns in primary care with a novel temporal phenotyping approach under uncertainty: A retrospective cohort analysis in UK

Abstract submitted to Healthcare Engineering at UCL ECR Symposium 2021, Sep 2021

  • Data frameworkfor measuring multi-morbidity across different locations: the challenge for the Health Data Research (HDR) UK MM Implementation Project

Project Team and Collaborators

  • Daniel Alexander
  • Spiros Denaxas
  • Eda Ozyigit
  • Arturo Gonzalez-Izquierdo
  • Muhammad Qummer Ul Arfeen


Health Informatics, Multimorbidity, Electronic Health Records, Disease Trajectories