Context:
A healthcare client conducting global clinical trials struggled to identify patients at high risk of stroke due to scattered medical records and inconsistent manual assessments.
Resolution:
We deployed a team of experts and used R to build a predictive model with tidyverse, caret, and random Forest. Data was cleaned, features were engineered, and a risk scoring model was created.
Result:
The model achieved 85 percent accuracy, enabling early intervention. High risk patients were flagged automatically, reducing emergency stroke cases by 20 percent.