Extreme value analysis has the potential to improve spirometry interpretation

Area of Science:

• Pulmonary Medicine
• Biostatistics
• Respiratory Physiology

Background:

• Spirometry is crucial for diagnosing and managing respiratory diseases.
• Interpretation often assumes healthy lung function follows a Gaussian distribution.
• This assumption may not accurately reflect impaired lung function.

Purpose of the Study:

• To investigate if lung function impairment follows a non-Gaussian distribution.
• To apply extreme value analysis for modeling impaired lung function.
• To develop a more objective method for interpreting spirometry results.

Main Methods:

• Hypothesized non-Gaussian distribution for impaired lung function.
• Utilized Gumbel distribution to model lung function impairment.
• Calculated relative probability comparing healthy (Gaussian) and impaired (Gumbel) distributions.

Main Results:

• Demonstrated the utility of relative probability in simulated cases.
• Provided objective delineation of the uncertainty zone between normal and impaired function.
• Showcased a more analytic assessment of lung function impairment.

Conclusions:

• Treating impaired lung function as a separate distribution improves assessment.
• Extreme value analysis and relative probability offer objective discrimination.
• A more precise definition of the uncertainty zone enhances interpretation accuracy.