The team created a prediction algorithm that estimated the risk of patient discontinuation based on several factors, including physician and patient profiles, dose modification behavior, and fulfillment trends observed in the specialty pharmacy data.
ProcDNA developed logistic regression and decision tree regression models using historical data, with patient discontinuation as the dependent variable and various influencing factors as independent variables.
A risk score system (ranging from 1 to 5, with 5 representing high risk) was developed for all patients in the specialty pharmacy data. These scores were shared with the field teams to provide actionable insights.
The prediction model provided real-time guidance to the field teams, enabling them to prioritize physicians with high-risk patients for discontinuation, thereby improving patient management.
The model resulted in a 14% reduction in patient discontinuations, significantly enhancing patient adherence and improving overall business outcomes.
