Development of an Intelligent Clinical Decision Support System for Spirometry Quality Control

Area of Science:

• Pulmonary Medicine
• Artificial Intelligence
• Medical Diagnostics

Background:

• Spirometry is crucial for diagnosing respiratory diseases but relies on manual quality control.
• Current visual inspection methods are subjective and time-consuming.
• Accurate spirometry maneuver execution is vital for reliable test results.

Purpose of the Study:

• To develop an intelligent clinical decision support system for automated spirometry quality control.
• To enhance the accuracy and efficiency of spirometry test assessments.
• To integrate patient demographic data into the quality control process.

Main Methods:

• A convolutional neural network (ResNet-18) was employed for analysis.
• The system integrates spirometry flow-volume and volume-time curves with patient demographics (sex, age, BMI).
• A graphical construct was generated for input into the neural network.

Main Results:

• The system achieved a promising Area Under the Curve (AUC) of 0.94.
• Sensitivity was 75.00% and specificity was 100.00% at the selected cut-off.
• The AI model effectively quantifies the risk of unacceptable spirometry tests.

Conclusions:

• The developed AI system shows potential for improving spirometry quality control.
• Further validation in real clinical settings with diverse datasets is necessary.
• The system requires broader implementation studies to assess robustness and generalization.