Nilakash Das is PhD fellow at the Laboratory of Respiratory Diseases and Thoracic Surgery (BREATHE) of KU Leuven. His work focuses on the clinical application of data sciences and artificial intelligence in the respiratory field. For his most recent work he received the ERS Young Scientist award.
The paper focuses on the development and validation of an artificial intelligence-based method for standardization of ATS/ERS quality criteria for spirometry to determine acceptability and usability results. In this interview he shares more about this work.
WHAT WAS THE MOTIVATION TO DEVELOP A METHOD FOR SPIROMETRY QUALITY CONTROL?
While ATS/ERS standards for spirometry quality include several quantitative limits, it also requires a technician to visually inspect flow-volume curves. This approach is time consuming and leads to a high inter-technician variability. In-fact, during my own research on spirometric biomarkers, I spent a lot of time manually examining thousands of curves to ensure quality of the data. Therefore, I wanted to develop a computer vision algorithm that takes into account the visual cognition of skilled technicians and quantitative guidelines, in determining acceptability and usability of a spirometric manoeuvre.
WHAT ARE THE BENEFITS OF THIS METHOD?
The AI based quality assurance algorithm provides a consistent interpretation on manoeuvre acceptability, and minimizes inter-technician variability. It also has the potential to provide manoeuvre feedback in real-time.
IN SEPTEMBER 2019, NEW ATS/ERS CRITERIA WERE PUBLISHED. HOW DID THAT AFFECT YOUR WORK?
The 2019 standards provide a lot of clarity on acceptability and usability of manoeuvres. We are actively working on updating our algorithm, which was based on 2005 standards. A key challenge lies in validation, as we do not have any data that has been evaluated with the latest standards. Therefore, we are also exploring ways to validate it in a clinical setting.
WHat future do you see for the developed method?
Our application can assist technicians and primary care physicians in daily practice to comply better with ATS/ERS quality criteria for spirometry while performing the test. It can also be used as an independent module in retrospective evaluation of spirometry curves e.g. in clinical and epidemiological studies. In the post-COVID19 era where home monitoring is gaining mainstream adoption, our system can remotely verify acceptability requirements, and further provide feedback to the user.