Innovative Approach in Nursing Care: Artificial Intelligence-Assisted Incentive Spirometry

Area of Science:

• Artificial Intelligence in Healthcare
• Digital Health Technologies
• Respiratory Medicine

Background:

• Incentive spirometry is crucial for pulmonary function and reducing postoperative complications.
• Traditional spirometers lack real-time feedback and automated monitoring capabilities.
• Manual nursing workload for respiratory exercises can be significant.

Purpose of the Study:

• To explore the feasibility of an AI-supported system for automating incentive spirometry monitoring.
• To reduce manual nursing workload through automated respiratory exercise tracking.
• To enhance patient adherence and outcomes in respiratory care.

Main Methods:

• Utilized a tablet camera for real-time tracking of a standard spirometer's volume indicator.
• Employed image processing techniques for analyzing exercise performance.
• Trained machine learning models (Random Forest, XGBoost, etc.) on patient data to classify respiratory performance.

Main Results:

• AI system achieved 100% accuracy and R² = 1.0 with ensemble methods.
• Cross-validation mean accuracies for machine learning models exceeded 99.4%.
• Demonstrated technical viability for AI-driven respiratory monitoring.

Conclusions:

• The AI-supported system is a cost-effective and accessible solution for clinical and home settings.
• Potential to integrate into standard respiratory care protocols, reducing nursing workload.
• Shows promise for improving patient adherence and outcomes in respiratory rehabilitation.