Using Machine Learning to Improve the Prediction of Functional Outcome in Ischemic Stroke Patients
See where this sits in the topic map →Summary AI-generated
- TL;DR
- Ischemic stroke is a leading cause of adult disability and death worldwide, and patient prognosis depends heavily on acute-phase treatment decisions.
- Problem
- Predicting a patient's functional outcome three months after an ischemic stroke, building on existing rule-based admission scores like ASTRAL, DRAGON, and THRIVE.
- Method
- Applying machine learning techniques to patient data and progressively incorporating features available at later points in time.
- Results
- Using admission features alone, a pure machine learning approach achieves an AUC of 0.808±0.085, which is only marginally superior to the best existing rule-based score at 0.771±0.056. However, progressively adding features from further points in time significantly increases the AUC above 0.90.
- Contributions
- Not specified in the abstract.
- Limitations
- Not specified in the abstract.
- Takeaways
- The findings validate the usefulness of current rule-based scores at the time of admission while highlighting the value of incorporating broader features over time using more advanced methods.
- Applications
- Not specified in the abstract.
- Topics
- Machine Learning; Healthcare; Stroke Prognosis; Clinical Decision Support
- For industry
- Healthcare and medical services
- Why it matters
- Not specified in the abstract.
Abstract
Ischemic stroke is a leading cause of disability and death worldwide among adults. The individual prognosis after stroke is extremely dependent on treatment decisions physicians take during the acute phase. In the last five years, several scores such as the ASTRAL, DRAGON, and THRIVE have been proposed as tools to help physicians predict the patient functional outcome after a stroke. These scores are rule-based classifiers that use features available when the patient is admitted to the emergency room. In this paper, we apply machine learning techniques to the problem of predicting the functional outcome of ischemic stroke patients, three months after admission. We show that a pure machine learning approach achieves only a marginally superior Area Under the ROC Curve (AUC) ( 0.808±0.085) than that of the best score ( 0.771±0.056) when using the features available at admission. However, we observed that by progressively adding features available at further points in time, we can significantly increase the AUC to a value above 0.90. We conclude that the results obtained validate the use of the scores at the time of admission, but also point to the importance of using more features, which require more advanced methods, when possible.