A carregar...
Publicação
Interpreting Stroke Prognosis: Key Factors Influencing Modified Ranking Scores at Discharge and Beyond
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 591.34 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Stroke is a leading cause of disability and mortality worldwide. Accurately predicting patient
outcomes, specifically using the modified Rankin Scale (mRS), is crucial for tailoring
rehabilitation strategies and allocating healthcare resources efficiently. However, the
complexity of stroke pathophysiology and the variability in patient responses make this a
challenging task. This study investigates the factors driving the severity of mRS scores at
discharge and three months post-discharge.
Four predictive models were employed, with SHAP (Shapley Additive Explanations) used to
interpret factor significance. Our findings indicate that the XGBoost (Extreme Gradient
Boosting) model outperformed other algorithms, achieving an Area Under the Curve (AUC) of
79% at discharge and 87% three months after discharge.
The analysis revealed a dynamic change in factor significance over time. The NIHSS (National
Institutes of Health Stroke Scale) was the most critical factor at discharge, while the patient's
post-discharge destination became more significant at three months. This shift highlights the
evolving nature of stroke recovery and the necessity for time-sensitive interventions.
This study contributes to understanding stroke recovery by identifying key influencing factors
and demonstrating the utility of SHAP values for interpreting more complex models. These
insights highlight the importance of prompt intervention and thorough analysis of patient
data, informing the development of precise and interpretable predictive models. The findings
have practical implications for healthcare practitioners aiming to improve patient
management and treatment strategies in clinical practice.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Knowledge Management and Business Intelligence
Palavras-chave
healthcare stroke modified ranking scale machine learning interpretable classification outcome prediction SDG 3 - Good health and well-being