Comparison of support vector machine based on genetic algorithm with logistic regression to diagnose obstructive...

Area of Science:

• Medical informatics
• Data mining in healthcare
• Sleep medicine

Background:

• Obstructive sleep apnea (OSA) diagnosis is crucial in medicine.
• Polysomnography (PSG) is the gold standard but resource-intensive.
• Data mining offers potential alternatives for OSA diagnosis.

Purpose of the Study:

• To compare the diagnostic performance of Support Vector Machine (SVM) and Logistic Regression (LR) models for Obstructive Sleep Apnea (OSA).
• To evaluate the feasibility of using data mining techniques as a substitute for Polysomnography (PSG).

Main Methods:

• A cohort of 250 patients diagnosed with OSA via PSG was analyzed.
• Logistic Regression (LR) model selection utilized Akaike's Information Criterion (AIC).
• Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel, optimized by a genetic algorithm, was employed.

Main Results:

• The best LR model included all variables, as determined by AIC.
• LR model achieved an accuracy of 0.797, sensitivity of 0.714, and specificity of 0.847.
• SVM model achieved an accuracy of 0.729, sensitivity of 0.777, and specificity of 0.702.

Conclusions:

• Both LR and SVM models demonstrated appropriate performance in diagnosing OSA.
• Considering accuracy, the SVM model may offer better efficiency than LR for OSA diagnosis.
• Further research could validate SVM's role in clinical OSA diagnosis.