The Prediction Models for High-Risk Population of Stroke Based on Logistic Regressive Analysis and Lightgbm Algorithm...

Area of Science:

• Epidemiology
• Medical Informatics

Background:

• Stroke remains a leading cause of disability and mortality worldwide.
• Identifying high-risk populations is crucial for effective stroke prevention strategies.

Purpose of the Study:

• To investigate high-risk factors for stroke using logistic regression and the LightGBM algorithm.
• To compare the predictive performance of these two models for stroke risk.

Main Methods:

• A cohort of 2124 residents over 40 in Jiaxing, China, was analyzed.
• Logistic regression and LightGBM algorithms were employed to build stroke risk prediction models.
• Model performance was evaluated using F1 score, accuracy, recall, and AUROC.

Main Results:

• Key stroke risk factors identified include age, male gender, high-salt diet, and alcohol consumption frequency.
• Negative correlations were observed with fruit consumption frequency and education level in older adults.
• The LightGBM algorithm demonstrated superior accuracy and AUROC compared to logistic regression.

Conclusions:

• Age, gender, diet, alcohol consumption, and education level are significant stroke risk predictors.
• The LightGBM algorithm offers a more accurate approach for identifying individuals at high risk of stroke.
• Findings can inform targeted interventions for stroke prevention.