Artificial Intelligence (AI) to analyse patient stories in the context of Person-Centred Nursing quality: the nurses’ perspective

Background and Objectives
To provide a new perspective on nursing quality eight Key Performance Indicators (KPI) and an associated Measurement Framework
were developed, with patient stories as a central data source. To improve data collection the iMPACT-App was developed, where the
stories are automatically transcribed. They are categorised and tagged to the KIPs by nurses. This is time consuming and is often
skipped due to time constraints. The objective is to automate the procedure by using AI.
Design and Methods
In order to gain insight into the criteria for an AI system capable of semantic search and text mining in collaboration with users, focus
groups were conducted with nurses. The data collected was analysed through reflexive thematic analysis.
Results
Six themes were developed, covering accessibility, usability, boundaries, safety, control of a human over the AI, and result
representation. It was important to the participants to enable all patients on the ward to participate in data collection and provide an
easy-to-use solution, which takes the needs of their patients into account. Given the desire for a high level of automation and the trust in
AI reported by participants, it was crucial to ensure the presence of a human in the loop, specifically the nurse and the patient.
Anonymising the stories and obtaining informed consent were discussed from the patient and the nurse perspective, to ensure patient
safety and create a psychological safe work environment. A key point regarding patient safety was the potential of using AI to identify
traumatic incidents in stories, in this context further discussed were organisational responsibilities.
Conclusions
The participants demonstrated a clear intention to develop an inclusive AI system to facilitate engagement with patients, who were
typically excluded due to barriers. Furthermore, a discussion concerning the ethical responsibilities regarding the development of AI is
essential. AI displays both potential and limitations in the analysis of qualitative data, such as stories, with challenges that are analogous
to those encountered in qualitative data analysis conducted by humans. Moreover, the insights gleaned are not confined to AI alone, as
it was considered as part of the broader context of the app, and thus could not be isolated from it