Insights into spirometry usage and the potential for AI-based decision support software in Belgian primary care

Read the full abstract here.

Objective: The first goal is to understand the current delivery of and barriers to performing spirometry in Belgian primary care settings. Second, we investigate the added value of AI-based software (ArtiQ.Spiro) for the quality assessment and interpretation of curves. Third, by focusing on a population at risk of COPD without formal diagnosis, we study how ArtiQ.Spiro can support general practitioners (GPs) in diagnosing COPD.

Methods: Structured interviews were conducted in 18 Belgian GP practices at baseline and at the end of the study. GPs were asked to primarily include patients for doing spirometry according to following criteria: patients older than 35 years old, with a smoking history ≥ 10 pack-years and at least one respiratory complaint revealed in the anamnesis. ArtiQ.Spiro automatically assesses quality and acceptance of the curves and suggests a diagnosis. After every use, the added value of the software was rated on a Likert scale (1-5).

Results: Questionnaires revealed that spirometry is important in primary care, but it is currently not used according to the latest guidelines. Barriers relate to lack of good training and time pressure. 28.5% of patients at risk of COPD were flagged as having COPD. ArtiQ.Spiro was useful for both quality assessment (4.13/5) and diagnostic support (4.01/5).

Conclusion: Spirometry in a primary care setting has an important place to reduce underdiagnosis of COPD patients. An AI-powered decision support software may dramatically increase the consistent use of spirometry in primary care, and assures it is used in line with current guidelines.

Authors:

M. De Vos, K. Van Orshoven, P. Desbordes, J.Maes, V.Vanheule, C. Quataert, I.Fovel, M. Topalovic