Novel approach to computerized breath detection in lung function diagnostics

Area of Science:

• Pulmonary Medicine
• Biomedical Engineering
• Signal Processing

Background:

• Accurate breath detection is critical for lung function analysis, but current algorithms struggle with poor data quality, especially in infants.
• Existing threshold or smoothing algorithms lack accuracy in suboptimal data conditions, hindering reliable clinical index calculation.
• Precise breath end localization is paramount for meaningful interpretation of respiratory data.

Purpose of the Study:

• To develop a novel algorithm for accurate breath detection in severely distorted respiratory data.
• To address the limitations of current methods in handling suboptimal data quality, particularly for infant lung function analysis.
• To create a universal and robust breath detection solution applicable across various patient groups and breathing disorders.

Main Methods:

• Utilized flow and gas concentration data from multiple breath washout tests as input.
• Designed a novel breath detection algorithm based on universal physiological characteristics of the respiratory tract.
• Validated the algorithm's accuracy on severely distorted data from 19 patients with diverse breathing disorders, comparing it to existing algorithms and expert estimations.

Main Results:

• The novel algorithm demonstrated superior accuracy compared to traditional threshold algorithms.
• Its performance in detecting breaths was comparable to that of human experts.
• The algorithm proved highly robust and efficient, even with severely distorted data from patients with various pulmonary conditions.

Conclusions:

• The proposed algorithm is highly robust, universal, and accurate, even in severely distorted data where other methods fail.
• It eliminates the need for pre-set thresholds or patient-specific inputs, allowing for broad patient applicability.
• This algorithm has the potential to supersede current breath detection approaches in pulmonary function diagnostics.