"Help Humans Make the Right Decisions.” Patient Perspectives on Artificial Intelligence (AI) Supporting Clinicians in the Interpretation of Spirometry: A Qualitative Study

Abstract

Abstract Rationale: Quality and accuracy of interpretation of spirometry performed in primary care is variable. Artificial Intelligence (AI) support software has been shown to be of benefit to pulmonologists interpreting lung function tests and has recently been validated in primary care datasets with potential to support primary care clinicians with spirometry interpretation. We aimed to understand patient perspectives on AI decision support software in aiding clinicians to perform and interpret spirometry. Methods: Participants enrolled in a real-world evaluation of an AI support software (ArtiQ.Spiro) in Primary Care spirometry pathways were invited to take part in a focus group to explore the acceptability of the intervention and patient perspectives on AI in this setting. The focus group was conducted on-line via a video conference call, recorded and transcribed verbatim. Two qualitative researchers facilitated the focus group and a topic guide was used to direct the discussion to capture key information. A structured framework approach was used for analysis based on previous qualitative themes relating to this programme of work1 and specifically to understand the patient view on their clinician receiving an AI report and the perceived values or challenges of using AI reports for spirometry. Results: Nine participants undergoing spirometry in primary care and enrolled in the real-world study completed the focus group: 5 female, 4 male, age range: 44 – 81 years. Themes developed from the data were 1) AI can help clinicians make better decisions, and 2) the human element should not be lost from clinical care (Figure 1). Participants described potential benefits in speeding up the process for their spirometry results, whilst expressing some concerns that a clinician ought to retain oversight of the spirometry report and diagnostic outcomes. Conclusion: Overall participants expressed views that AI was a positive addition to healthcare but that human interaction was still important and they valued clinician ability to query the AI outcome. 1. Doe, G. et al. Spirometry services in England post-pandemic and the potential role of AI support software: a qualitative study of challenges and opportunities. British Journal of General Practice73, e915-e923, doi:10.3399/bjgp.2022.0608 (2023).

Ähnliche Arbeiten