A novel AI-based diagnostic model for pertussis pneumonia

Area of Science:

• Medical Diagnostics
• Infectious Diseases
• Machine Learning in Medicine

Background:

• Clinical diagnosis of pertussis remains difficult, often relying on subjective physician experience.
• Accurate and timely diagnosis is crucial for effective patient management and public health control of pertussis.

Purpose of the Study:

• To develop and evaluate a machine learning-based diagnostic model for pertussis using biochemical blood test parameters.
• To compare the performance of different machine learning algorithms in diagnosing pertussis.

Main Methods:

• A retrospective study included 590 patients (295 pertussis, 295 non-pertussis lower respiratory infections).
• Univariate logistic regression identified significant clinical and biochemical features.
• Diagnostic models were built using K-nearest neighbor, support vector machine, and eXtreme Gradient Boosting (XGBoost) algorithms.

Main Results:

• Eighteen of 27 features were identified as optimal predictors for pertussis.
• The XGBoost model demonstrated superior performance compared to support vector machine and K-nearest neighbor models.
• The XGBoost model achieved an area under the receiver operating characteristic curve of 0.96 and an accuracy of 0.923.

Conclusions:

• A diagnostic model integrating blood biochemical test results with the XGBoost algorithm offers a promising tool for accurate pertussis diagnosis.
• This approach can significantly assist healthcare professionals in the effective diagnosis of pertussis.