Monitoring Lung Mechanics during Mechanical Ventilation using Machine Learning Algorithms

Area of Science:

• Biomedical Engineering
• Computational Physiology
• Artificial Intelligence in Medicine

Background:

• Accurate assessment of lung mechanics is crucial for optimizing mechanical ventilation strategies.
• Current methods for evaluating respiratory resistance (R) and compliance (C) may not be suitable for real-time, bedside application.
• Developing advanced computational models is key to personalized healthcare and context-aware interventions.

Purpose of the Study:

• To present a novel machine learning approach for the continuous, real-time bedside assessment of respiratory resistance (R) and compliance (C).
• To validate the performance of machine learning algorithms in estimating R and C using experimental data from a simulated lung model.
• To explore the potential of these algorithms in advancing next-generation ventilation technologies.

Main Methods:

• Development of supervised machine learning algorithms (decision tree, decision table, Support Vector Machine) to estimate R and C from flow rate and airway pressure data.
• Conducting an experimental study using a pressure control ventilator connected to a test lung simulating various R and C values.
• Validation of algorithms using collected sensor data and comparison with a linear regression model.

Main Results:

• Machine learning algorithms demonstrated high accuracy in assessing respiratory R and C: decision table (90.3%), decision tree (93.1%), and SVM (63.9%).
• A linear regression model achieved an impressive 99.4% accuracy in estimating R and C.
• The study confirms the feasibility of real-time R and C estimation using computational models.

Conclusions:

• Machine learning offers a promising approach for accurate and continuous bedside assessment of lung mechanics (R and C).
• The developed algorithms have the potential to enable personalized ventilation strategies and context-aware interventions.
• This work paves the way for a new generation of intelligent ventilation technologies.