Post-stroke Anxiety Analysis via Machine Learning Methods.

Area of Science:

• Neurology
• Psychiatry
• Medical Informatics

Background:

• Post-stroke anxiety (PSA) is a significant concern with ongoing research into its risk factors and prediction.
• Machine learning (ML) offers novel approaches for analyzing complex health data and improving clinical outcomes.

Purpose of the Study:

• To identify risk and protective factors for PSA in acute ischemic stroke patients.
• To develop and evaluate ML models for predicting PSA.

Main Methods:

• Collected data from 395 acute ischemic stroke patients, assessed using anxiety scales (HADS-A, HAMA, SAS).
• Compared demographic and laboratory data between anxiety and non-anxiety groups to find statistically significant differences.
• Utilized multivariate logistic regression and various ML methods (including random forest) for risk factor analysis and PSA prediction.

Main Results:

• Hypertension, diabetes mellitus, drinking, high NIHSS scores, and low serum HDL-C levels were identified as significant risk factors for PSA.
• Higher serum HDL-C levels demonstrated a protective effect against PSA.
• The random forest model showed superior performance in predicting PSA compared to other ML methods.

Conclusions:

• Hypertension, diabetes mellitus, drinking, and disability are key risk factors for PSA, while higher HDL-C is a protective factor.
• Machine learning models, especially random forest, can effectively predict PSA, aiding in early clinical intervention strategies.