Using machine learning models to classify stroke risk level based on national screening data^.

Area of Science:

• Public Health
• Medical Informatics
• Cardiovascular Disease Research

Background:

• Stroke poses a significant public health challenge in China due to its high incidence, prevalence, and mortality.
• The national stroke screening program faces limitations in classifying high-risk individuals when stroke risk factors are unknown, impacting intervention efficiency and statistical accuracy.

Purpose of the Study:

• To develop and evaluate machine learning models for accurate stroke risk classification, addressing data gaps in the national screening program.
• To improve the efficiency of stroke interventions and reduce associated healthcare expenditures.

Main Methods:

• Data preprocessing, including handling imbalanced datasets using oversampling and undersampling techniques.
• Development and comparison of logistic regression, decision tree, neural network, and random forest models for stroke risk classification.
• Model evaluation based on precision and recall metrics.

Main Results:

• The random forest model exhibited the best performance in terms of recall and precision for stroke risk classification.
• The developed models effectively address the challenge of unknown risk factor values, improving classification accuracy.

Conclusions:

• Machine learning models, particularly the random forest approach, can significantly enhance the accuracy and efficiency of stroke risk screening.
• These models offer a valuable tool for optimizing public health interventions and resource allocation in stroke prevention programs.