Development of an AI Model for Predicting Methacholine Bronchial Provocation Test Results Using Spirometry.

Area of Science:

• Pulmonary Medicine
• Artificial Intelligence in Healthcare
• Diagnostic Technologies

Background:

• Methacholine bronchial provocation test (MBPT) is crucial for diagnosing asthma and airway hyper-reactivity.
• MBPT is often time-consuming and resource-intensive, necessitating alternative diagnostic methods.
• Spirometry, including forced expiratory volume in one second (FEV1) and bronchodilator response, provides readily available data.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for predicting MBPT outcomes.
• To utilize standard spirometry measurements (FEV1, bronchodilator response) as input for the AI model.
• To assess the potential of AI in streamlining asthma diagnostic workflows.

Main Methods:

• A dataset of spirometry measurements, including pre- and post-bronchodilator FEV1, was compiled.
• Various machine learning models, including Multilayer Perceptron (MLP), Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), and Extreme Gradient Boost (XGBoost), were trained and validated.
• Feature importance analysis was conducted on the best-performing model to identify key predictive factors.

Main Results:

• The MLP model demonstrated the highest performance with an Area Under the Curve (AUC) of 0.701, accuracy of 0.758, and F1-score of 0.853.
• LR and SVM models showed comparable AUC values around 0.688.
• Key features influencing predictions included Pre-FEF25-75 (%), Pre-FVC (L), Post FEV1/FVC, Change-FEV1 (L), and Change-FEF25-75 (%).

Conclusions:

• AI models can effectively predict MBPT results using standard spirometry data, particularly FEV1 and bronchodilator responses.
• This AI-driven approach has the potential to enhance the efficiency and accessibility of asthma diagnosis.
• The findings support the integration of AI into clinical practice for improved diagnostic workflows and reduced reliance on MBPT.