Machine learning models to identify low adherence to influenza vaccination among Korean adults with cardiovascular...

Area of Science:

• Public Health
• Cardiovascular Disease (CVD)
• Vaccination Adherence
• Machine Learning

Background:

• Annual influenza vaccination is crucial for cardiovascular disease (CVD) patients, particularly during the COVID-19 pandemic.
• Identifying CVD patients with low influenza vaccination adherence is essential for public health strategies.
• This study focuses on the Korean adult population using a nationally representative health dataset.

Purpose of the Study:

• To develop and compare machine learning models for identifying Korean adult CVD patients with low influenza vaccination adherence.
• To analyze vaccination patterns across different age groups, considering free immunization programs for the elderly.

Main Methods:

• Utilized data from the Fifth Korea National Health and Nutrition Examination Survey (KNHANES V) for 815 adult CVD patients.
• Employed four machine learning techniques: logistic regression (LR), random forest (RF), support vector machine (SVM), and extreme gradient boosting (XGB).
• Developed separate models for age groups under 65 and 65 and older due to differing vaccination policies.

Main Results:

• For individuals aged 65 and older, XGB and RF models achieved the highest accuracy (84.7%) in predicting low vaccination adherence.
• For individuals under 65, the SVM model demonstrated the best performance with 68.4% accuracy.
• The study compared the predictive power of 16 variables across the different machine learning models.

Conclusions:

• Machine learning models demonstrate comparable effectiveness in classifying adult CVD patients with low adherence to influenza vaccination.
• The findings can inform targeted interventions to improve influenza vaccination rates among vulnerable CVD patient populations in Korea.