See https://github.com/joshyaffee/healthcare_stroke_ML/blob/main/StrokePredictionML.ipynb for our code and work description.
Strokes are a serious medical condition that occur when the blood supply to the brain is disrupted, either by a blood clot or a ruptured blood vessel. Strokes can cause significant disability or even death, and early detection and prevention is critical for improving patient outcomes.
Predicting strokes is important in healthcare for several reasons. First, strokes are a leading cause of death and disability worldwide, and early detection and intervention can help reduce the risk of stroke and improve patient outcomes. Second, predicting strokes can help identify patients who are at higher risk of stroke and enable healthcare providers to provide preventative measures, such as medication or lifestyle changes, to reduce that risk. Third, predicting strokes can also help healthcare providers allocate resources more efficiently, by identifying patients who are most likely to benefit from additional screening or intervention.
Common tools used to predict strokes include traditional risk assessment tools, such as the Framingham risk score, which use demographic and clinical factors to estimate a patient's risk of developing cardiovascular disease, including stroke. More recently, machine learning algorithms have been developed that can predict stroke risk using a wider range of factors, such as genetic and lifestyle factors, and can provide more personalized risk assessments.
Predicting strokes can be challenging for doctors, as the risk factors for stroke are complex and multifactorial. In addition, stroke risk can be influenced by a wide range of factors, including age, sex, lifestyle, genetics, and medical history. As a result, traditional risk assessment tools may not always provide accurate predictions, and doctors may not have access to all of the relevant information needed to make a reliable prediction. Machine learning algorithms can help address these challenges by analyzing large amounts of data and identifying patterns and risk factors associated with stroke.
We will employ the following analysis strategy for using machine learning to predict strokes.
- Evaluate dataset features
- Oversample the dataset
- Attempt several models
- Ensemble of models
- Validation