Developed a model to predict patient discontinuation for an oral drug using specialty pharmacy data
Business Problem
Client wanted to improve patient adherence. They wanted to use the rich specialty pharmacy data to provide specific and actionable insight to the field teams to reduce patient discontinuations.
Our Approach
Prediction Algorithm
The ProcDNA team developed a prediction algorithm to estimate the risk of patient discontinuation based on several factors such as physician and patient profile, physician dose modification behavior, fulfilment trends seen in specialty pharmacy data
Decision Tree
We developed logistic and decision tree regression models using the number of patients discontinued (dependent variable) and different independent variables
Risk Score
Finally, we developed a risk score “1-5” (5 = high risk) for all patients in specialty pharmacy data, shared scores with field teams
Client Impact
Real-time Guidance
Provide real-time guidance to field teams on physicians with high-risk patients for discontinuation
Measurable Success
Reduced patient discontinuation by 14%