Analysis of the Cardiorespiratory Pattern of Patients Undergoing Weaning Using Artificial Intelligence.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Critical Care Medicine

Background:

• Determining the optimal moment for extubation in mechanically ventilated patients remains a clinical challenge.
• Respiratory pattern variability analysis offers a potential method to aid in this decision-making process.

Purpose of the Study:

• To analyze respiratory pattern variability using artificial intelligence techniques.
• To identify the optimal extubation moment by classifying patients into successful, failure, and reintubated groups.

Main Methods:

• Analysis of respiratory flow and electrocardiogram time series signals.
• Application of Power Spectral Density, Discrete Wavelet Transform, and time-frequency domain analysis.
• Development of a Q index for parameter selection and dimensionality reduction using forward selection and bidirectional techniques.
• Classification of patients using Linear Discriminant Analysis and Neural Networks.

Main Results:

• High classification accuracies were achieved: 84.61% (successful vs. failure), 86.90% (successful vs. reintubated), and 91.62% (failure vs. reintubated).
• Parameters derived from the Q index and Neural Networks demonstrated superior performance in patient classification.

Conclusions:

• Artificial intelligence-based analysis of respiratory pattern variability can effectively aid in determining the optimal extubation time.
• The proposed Q index and Neural Networks offer a promising approach for predicting extubation outcomes in mechanically ventilated patients.