Characterising the dynamic inter-relationships between polypharmacy and multiple long-term conditions. Using artificial intelligence (AI) to map patient journeys into multimorbidity clusters across the UK

Overarching question: “How do polypharmacy and multiple long-term conditions (multimorbidity)(MLTC-M) interact and can enhanced understanding of their inter-relationships help to define optimised treatment strategies?” Overarching aim: to better understand the complex interactions between polypharmacy and disease clustering over the life-course.

Our ultimate goal is to identify tipping points associated with progressive multimorbidity and polypharmacy which could trigger pre-emptive interventions.

Appropriately designed clinical decision support systems, including alerts for electronic health records (EHRs), would then inform patient-centered conversations around management options.

Background MLTC-M is associated with premature mortality, significant treatment burden, increased healthcare and socio-economic costs. Factors contributing to a trajectory of increasing multimorbidity remain to be defined.

Polypharmacy (defined as the simultaneous use of five or more medications) is associated with MLTC-M but their inter-relationship is poorly understood.

Whereas appropriate polypharmacy may alleviate and/or prevent disease burden, inappropriate polypharmacy may do the reverse.

Access to data in EHRs and application of AI enables detailed study of the relationships between polypharmacy and MLTC-M over time, and in the context of other potential risk and protective factors.

Aims and objectives of Development Work Establish a functional consortium including PPI groups, ensuring optimal membership and configuration.

Refine our data engineering, AI and machine learning methodology and develop working definitions of “polypharmacy”.

Perform an exploratory analysis to identify proof-of-principle clusters of MLTC-M associated with, and defined by links to polypharmacy.

Methods Building on our combined expertise in innovative healthcare data science, during the Development Award we will: Establish data access, integration and sharing, across the two main centres, ensuring compliance with relevant governance standards.

Utilise clinical and prescription data within the UK Biobank to refine AI and machine learning methods and develop proof-of-principle MLTC-M/polypharmacy clustering outputs.

Establish access to anonymised EHR data in two geographically and demographically distinct, patient populations, in North East and North Cumbria and East London.

These datasets, representing ethnically and socially diverse populations will provide the basis for the full Research Collaboration application.

Impacts The contrasting ethnicity but similar deprivation across our two populations offers a unique opportunity to study MLTC-M and polypharmacy in the context of these important risk factors.

The Development Award will position us to investigate these two populations with clearly defined, PPI and stakeholder informed, research questions and methodologies.

During the full Research Collaboration, we will define detailed patient journeys into MLTC-M, providing a deeper understanding of the inter-relationships with polypharmacy and other drivers (e.g. Black, Asian and Minority Ethnic populations).

This will enable design of early warning and clinical decision support tools to flag risk to health professionals, pharmacists and patients.

Identification of potential 'tipping points will inform holistic, inclusive and patient-centered conversations around management options.

We will use target trial emulation to analyse longitudinal routine data and to optimally design prospective pragmatic trials, relevant to the management of MLTC-M. Dissemination Informed by consultation with our PPI groups, important findings and messages will be shared widely. A methodological paper outlining our analysis plans will be published in a clinically impactful peer reviewed journal.