JMIR Publications

ABSTRACT

Background:

Due to its association with multi-morbidity, frailty gives rise to multi-dimensional needs for different services. Too often, information about both patient preferences and service encounters is not adequately shared.

Objective:

This study aims to codesign, collect and analyze encounter data from multiple community/primary-based MDT services with people with frailty to develop prototype Large Language Models (LLMs) that can generate clinical AND person-centred care (PCC) summaries.

Methods:

Working with two primary care networks (PCN), we will engage stakeholders in the codesign of this research, ensuring that it is acceptable and relevant and meets local infrastructure, information governance, and regulation requirements. Audio recordings will be taken of at least three encounters between different members of MDTs and n=50 patients with frailty whom their General Practitioners (GPs) have identified as requiring MDT engagement. Recordings will be transcribed into text for concept design and model pre-training. We will combine this data with insights from the stakeholder engagement to understand key input prompts that will be required to develop sensitive AI models that respond to different stakeholders' needs, workflows, and preferences. To generate the person-centred summaries, we will test out two approaches to modelling the encounter data: graph-based modelling and hierarchical transformers. The AI-generated summaries will be compared to human written summaries of the same encounter data and assessed for accuracy, quality, fluency, and person-centredness. They will also be shared with original MDT members for validation. If deemed suitable for deployment, optimum ways of integrating these summaries into shared care records will be explored with local key system leaders.

Results:

No results to present.

Conclusions:

We will capture inputs, processes, and outcomes across all key phases of the implementation journey to identify capability requirements, determinants of implementation (inc. key challenges and best practices to overcome them) and the value added by the technology. Depending on the extent to which the model is deemed technologically ready, this information will be summarised in a guidance toolkit on 'how to' integrate person-centred (PC) summaries into shared care records.