Applying Hybrid Machine Learning Algorithm to Measure Total Lung Capacity from Spirometric Values

The accurate diagnosis of restrictive ventilatory impairment requires the measurement of Total Lung Capacity (TLC), a process that typically involves complex and resource-intensive techniques like plethysmography or gas dilution. However, these advanced methods are often unavailable in routine clinical settings, creating a need for alternative approaches that leverage more accessible tools, such as spirometry. This study proposes a machine learning- based system that estimates TLC directly from spirometry data using advanced tree-based models, aiming to streamline the diagnostic pathway for restrictive lung diseases. By accurately predicting TLC from standard spirometric metrics, this approach can identify patients who would most benefit from further pulmonary function testing (PFT), reducing the need for additional, costly testing procedures. Comparative analyses were conducted between the machine learning model and Conventional Spirometry-based restriction predictionpatterns, showing improved accuracy and reliability in identifying restrictive impairments. The proposed system demonstrates significant potential as a decision- support tool, enhancing clinical workflows by prioritizing patients for full PFTs and improving access to timely and accurate diagnosis of restrictive ventilatory impairments.