Clinical Scorecard: Predicting Early Neurological Deterioration in Acute Ischemic Stroke Using a Multidimensional Machine Learning Approach

At a Glance

Key Highlights

• Developed a machine learning model to predict early neurological deterioration (END) in AIS patients
• Identified five core predictors: NIHSS score, blood glucose, infarct core volume, collateral circulation status, NLR
• Random Forest model showed superior predictive performance with AUC values of 0.779 (training) and 0.775 (validation)
• END occurs in 10–40% of AIS patients and is linked to poor long-term outcomes
• The model aims to facilitate early identification and intervention for high-risk patients

Guideline-Based Recommendations

Diagnosis

• Utilize admission NIHSS score and blood glucose as initial assessment tools
• Consider imaging studies to evaluate infarct core volume and collateral circulation

Management

• Implement timely interventions for patients identified at high risk for END
• Monitor and manage systemic factors such as fever and dysglycemia

Monitoring & Follow-up

• Regularly assess NIHSS scores to detect any neurological deterioration
• Monitor blood glucose levels and other relevant clinical parameters

Risks

• Recognize that END is associated with prolonged hospitalization and increased mortality
• Understand that traditional predictors may not adequately capture the risk of END

Patient & Prescribing Data

338 AIS patients admitted within 24 hours of symptom onset

Focus on multidimensional assessment to improve risk stratification and outcomes

Clinical Best Practices

• Incorporate machine learning models in clinical decision-making for AIS
• Utilize a comprehensive approach to assess multiple clinical and pathological indicators
• Engage in multidisciplinary discussions to interpret predictive model outputs effectively

References

• Study on END in AIS