Predicting Cerebrovascular Diabetes Compilations with ML

Inspiration

Diabetes affects millions of people worldwide and around 90% of patients suffer from Type 2 diabetes. The effects of diabetes influence the entire body, including the brain, due to altered vasoregulation and blood flow in patients. However, biomarkers that allow scientists to predict cognitive decline and cerebral profusion in patients are still not available.

What it does

The XGBoost regression model uses patient history, blood data, and cerebral information to understand which symptoms are most heavily correlated with having diabetes.

How I built it

The dataset (https://doi.org/10.13026/whjz-e968) was pre-processed to remove values that the model can't understand, such as strings and booleans, and then was trained in Google Colab. The dataset was split into 75% training data, 25% testing data, and graphs displaying the correlation were created.

Challenges I ran into

• original dataset contained many null values, bools, had typos

Accomplishments that I'm proud of

• analyzing a new dataset that is relevant for many people

What I learned

• pre-processing datasets: normalizing, making sure the dataset is compatible with the model

What's next for Predicting Cerebrovascular Diabetes Compilations with ML

• Using different models
• More patient data

Built With

• colab
• https://doi.org/10.13026/whjz-e968
• python
• xgboost